Mohib Jafri

Building Seed Labs. BS + MS in EECS, Harvard.

Projects, writings, etc.


Hardware6
Writings7
Research4
Web32
Talks3
Venture4
Tesla3
Hackathons4
Software7
Shenanigans1
High School4
Dec 15, 2023
Research
Venture

Distributed Venture Firms

The next iteraton of venture capital

View paper here

Jul 22, 2022
Venture
Web3

Venture 3.0

A look ahead at why venture is going to change.

Early Stage Venture is Changing

We're on the cusp of a new wave in the early stage venture capital ecosystem. In particular, the rapid decentralization of the core pillars that make up early stage VC is happening right in front of us:

Access to capital is becoming democratized. Technology and globalization as an equalizing force have a second order effect of rapidly increasing the count of high net-worth individuals (HNWI). The founder/successful retail investor demographic is no longer flocking to become LPs in highly competitive funds— it is nearly impossible to be an LP at a Tier-1 VC when you're at competition with endowments, pension funds, etc. Instead, a more attractive option tends to be successful founders/investors running their own mini-funds (see Distributed Capital from the Halo team, Todd & Rahul's angel fund, the Logos Fund, among many more). And with the SEC's lowering the barrier on who qualifies as an accredited investor, it is clear that the micro-funds will become a norm.

Quality deal flow is a function of networks, and networks fundamentally belong to people, not brands. The fight to invest in the best new companies will be increasingly won by incredible people backed by incredible brands, not the opposite. At the onset, there was no A16Z without, well, “A” (Marc Andreessen) and “Z” (Ben Horowitz). Founders don't work with firms, they work with individuals (often GPs/Associates) at the firm. While a firm's brand building is important and relevant at later stages, the best deals come from personal connections, not from deals@vcfirm.com. Networks own deal flow, and people own networks.

Due diligence teams scale better when distributed. Larger firms with multiple funds can afford to hire research teams of PhDs to lead theses. This approach doesn't scale to cash-tight, often illiquid micro-funds. Instead, using the age-old wisdom of the crowd mechanism, crowdsourced due diligence can achieve comparable insights to research teams, assuming the crowd has the right background to assess a deal. Crowdsourced knowledge is especially relevant as many venture firms are downsizing their staff in reaction to the mid-2022 bear market (1 and 2).

Primitives of distributed due diligence exists today: the Hacker News community is a technical community that quickly sniffs out the good tools from the bad ones. With the right tooling, the future of due diligence will be commoditized as part of the growing service DAO space that is taking on the $263B consulting market. This is the natural next step to the specialization of labor: from generalists who will miss obvious basics, to specialists on staff paid per diem, finally to specialist groups compensated on an adhoc basis.

Major signs point towards early stage venture decentralizing on different dimensions. But why?

Coordination failure.

Coordination failure: an existential threat to innovation

Coordination failure describes suboptimal outcomes to games that players fail to cooperate in. A harmless example: when people crowd around airport baggage claim conveyors: a small group of people acted selfishly by approaching the conveyor belt to grab their bag quicker, blocking the view of the belt to others. For fear of missing one's own bag, everyone must then crowd the belt, resulting in more commotion and slower resolutions for everyone. While an extra few minutes to leave the airport isn't a matter of life or death, coordination failure largely explains the failure to achieve nuclear disarmament in the 20th century. Instead of countries spending money of positive, nation-nurturing activities after an agreement to phase out nuclear weapons programs, trillions were spent in fear that one country could defect and dominate others.

Early stage venture suffers from micro-fractures of coordination failure.

Intra vs. Inter Coordination Failure

The Ethereum community has worked hard on solving intra-coordination failure. Forefront thinkers in the space like Ameen Soleimani have crafted the lore of “slaying Moloch, the rationalist god of coordination failure,” designing smart contracts that protect against coordination failure within orgs (intra-coordination). Codified principles like minority protection, pro-rata rights, and the power to freely exit have significantly dropped the cost of iterating on governance structures. We pay deep respect to the MetaCartel Ventures community that brought these governance structures to the venture capital space, building one of the first venture-investing communities governed by smart contracts.

As solutions arise for intra-coordination within venture organizations, a higher-order problem presents itself: inter-coordination failure. What does a landscape in which thousands of traditional firms, equity crowdfunding groups, angels, etc. exist in imperfect competition with each other? How do we account for conflicts of interest in which there are overlapping affiliations of individuals and their respect networks?

Inter-coordination failure between venture organizations is an existential threat to innovation. At the pre/seed stage, valuations for hot companies tend to get pumped up by two archetypes: late-stage funds paradoxically writing small checks with no regards to their allocation/startup valuation as they solely seek better access to Series A/B rounds, or early-stage funds offering sky-high valuations to get preferential lead positions. The former blocks founders from the critical mentorship/value-adds that the best early-stage venture capitalists provide. The latter, if executed many times over, results in firms getting lackluster ownership (and therefore lackluster ROI) in potentially stellar companies. These inflated valuations come to the detriment of projects when raising a Series A/B implies a down-round: either existing investors must prop-up the valuation by any means to salvage an initial investment, or founders struggle to raise as down-rounds are red flags to potential new investors.

The affect of coordination failures like this in the venture space is catastrophic. Disjointed, poorly incentive-aligned systems like the current venture capital space are inherently set to run sub-optimally, even fail.

Venture 3.0: the future of venture capital

Venture 1.0 was about the ‘Sand Hill Road,' blockbuster institutional firm raising a mega funds to fund much of the development of internet. Because scaling technologies (namely, cloud infrastructure) were missing, many innovative companies poured cash into running and maintaining their own servers. This crutch, among the lack of other tools that made building ventures more costly, justified the need for big investments. Many of the rules of thumb of venture capital were written during this period.

Venture 2.0 was characterized by successful GPs and founders from Venture 1.0 making their own smaller funds and producing incredible returns (look no further than Chris Sacca's Fund I for Lowercase Capital, which is one of the best performing funds at 213x thanks to his early investments in Twitter, Uber, and Instagram). Following suite in leaner funds were the likes of equity crowdfunding (WeFunder, Republic) and the rise of the angels (AngelList).

Venture 3.0 describes the future of venture capital— a unified protocol linking the full spectrum of investment vehicles, from angels to firms. We propose a collaborative, positive-sum game with ownership and composability at the center of the protocol.

Venture 3.0 Begins with Aligning Community Incentives

We look to venture DAOs (decentralized autonomous organizations) as a design paradigm that presents itself, in its current form, as ultra-lightweight, token native venture investing groups. They do not prescribe decentralizing any particular aspect of the venture lifecycle. In fact, venture DAOs as a design paradigm are that which can be encompassed by any investing community: angel collectives, venture firms, and investment DAOs alike.

Some characteristics that make these vehicles special:

  • Low burn. Many traditional activities (deal flow management, accounting, due diligence) that would traditionally inflate overhead/staffing costs can be replicated with better coordination software (distributed due diligence, cap table management software, service DAO integration).
  • Cheap to start. The right software enables near-zero coordination cost, and legal costs are being amortized by legal-tech companies like Doola.
  • Community-centric. Well-designed tokens enable communities to engage in collective ownership and governance over both their funds and the organization's direction.
  • Scalable. Tokens can be designed to incentivize more people than just a “core” set of partners to engage in value-producing activities. Token incentives are a force multiplier for new and existing venture organizations to extend their reach far past those “on-staff.”

Venture DAOs are for everyone— from venture firms, to angel groups, to pseudonymous passion-groups. Composable infrastructure is key to the venture DAO paradigm adapting to a wide range of existing governance structures.

Note: Venture DAOs do not reinvent the wheel entirely. before venture DAOs were even a term, we've seen primitive nods to its inevitability: scout programs (Contrary, Open Scout, Draper, etc.), carry incentives, angel investment groups on AngelList, even equity crowdfunding sites like WeFunder have all taught the industry a lot about how early stage venture can be done differently. Venture DAOs are a design paradigm that enable us to take from these disparate spaces of learning, and craft new mechanisms better via composable tools.

Adopting the venture DAO paradigm also opens communities up to collaboration with the plethora of service DAOs that coordinate a distributed labor base for legal, accounting, design work, and more.

Scaling the Experiment

The Moloch and MetaCartel Ventures community served as incredible research points, proving that overcoming intra-coordination failure in venture is possible through codified governance.

Now, venture 3.0 must extend their work in two critical ways:

First, we must scaling intra-coordination success within venture DAOs, and extend venture DAO paradigm to existing entities in pursuit of community-first investing.

Second, we must develop solutions for inter-coordination issues to bring together hundreds of thousands of diverse entities in venture to interact in positive-sum games.

Collective Ownership is Necessary

Innovation is stuck in a loop. There is an ever-present ebb and flow— a bundling and unbundling— war between centralized parties and decentralized systems. New systems are made on open, decentralized protocols, corporations solve UX/aggregation problems and commoditize these protocols, community-wide anger with these rent-seeking parties disrupt the corporations with more open protocols, and the cycle repeats itself.

Collective prosperity is the only path forward. Much of institutional venture capital gets overturned by “disruptors” who then become the very institutions they sought to disrupt. Unnecessary competition like this (which damages the industry) comes from a lack of incentive alignment. The venture protocol of the future requires unification, not displacement, of all venture organizations via a well-designed token, one that enables all parties to share in the collective success of the new venture model.

Venture 3.0 does not displace the benefitted organizations in venture 1.0 and venture 2.0. It supercharges them, and enables the innovation-funding community at large to learn, iterate, and adapt together in ways we've never before seen possible.

Breathing New Life Into Venture

Nobody knows exactly what venture 3.0 will look like. In fact, venture 3.0 is characterized by scaling the ability to experiment with new mechanisms in venture.

We are at the dawn of a Cambrian explosion in venture. Better software, a favorable legal landscape, and an industry-wide urge for decentralization signals an unprecedented level of open experimentation that will breath new life into venture.

Seed Labs is powering the largest-scale experiment in early stage venture, with shared prosperity at the center of it.

You are a part of this story. Your contributions below— agreements and disagreements alike— will memorialize your beliefs in this turning point in venture.

This is not a product spec. This is not an advertisement.

It is a call for the venture world to experiment. A call to innovate. A call to imagine how we can do better.

Jun 1, 2022
Hardware

A Personal Carbon Capture Device

We built an at-home carbon capture device.

And it failed (successfully).

Writeup coming soon ⚡

Apr 2, 2022
Research
Web3

Optimal Delegation Strategies for DAOs

Wrote a paper in graduate school about decentralized autonomous organizations (DAOs) and how they can be used to optimize group decisionmaking.

A tweet that went off about it

Paper

Mar 11, 2022
Software

Red Cross x Help Aid Africa

Cooley Accredited Investor Questionnaire

Partnered with the Red Cross to launch the Help Aid Africa app, a logistics app that tracked physical donations from your doorstep in the US to the communities that receive the donation. I ran the team that designed, engineered, and launched the app to our end users.

The app tracked over 120,000 pounds of clothing, furniture, toys, and household products sent from California to Kenya.

Launch Video Here

Dec 15, 2021
Research
Software

Homomorphic Encryption Schemes for Federated Learning

View paper here

Sep 1, 2021
Software

A Better MIT Fall '21 Course Catalog

After coming back to the Harvard/MIT ecosystem for grad school, I grew impatient. I had two days left to register for MIT courses, and yet, none of the cool course-catalog-search-sites had updated to the Fall 2021 catalog. We were stuck with a really high-friction official MIT catalog. So I built a better one.

500+ customers ($3k profit) selling software + data for a burning problem to Harvard + MIT students selecting their Fall 2021 courseload. Ideated, shipped, and launched in < 24 hours.

Jun 1, 2021
Talks

DevPro 2021

Keynote speaker at one of Russia's largest developer conferences. Went over how my team at Tesla and at other startups approached building tech, with three main components:

  1. Validated learning + speed: implementing CI/CD organization-wide, not just in software engineering.
  2. Innovation management: how we nurture young and risky ideas but still turn a profit.
  3. Tooling: how Tesla makes project management tools work for teams — not the other way around.

Video

Jun 1, 2021
Talks

How Startups Build Tech: LeadxManage

I go over how my team at Tesla and at other startups approached building tech, with three main components:

  1. Validated learning + speed (continuous deployment + release MVPs constantly).

  2. Innovation management (how we nurture young and risky ideas but still turn a profit).

  3. Tooling: how Tesla makes project management tools work for teams — not the other way around.

Full talk available here

Mar 11, 2021
Software

A Personal CRM

Archive of product website here.

We launched this, got paying customers, then pivoted based on user feedback to messages.so.

Here is an old video before we changed the name + branding.

DM me to talk about why every personal CRM has failed (and why most current ones will probably fail too).

(And if you're building a better PRM, reach out. I would love to invest and be a user.)

Mar 11, 2021
Software

A Better GoFundMe

A better donation crowdfunding experience for the new generation— a sleek design, Apple Pay, share-to-story, and more. In less than three months, campaigns hosted on Seed raised $200k+ total.

Mar 7, 2021
Talks

How Startups Build Tech: LeadxManage

Big thanks to Execute Big for creating a brilliant program for high school students 🥂

Feb 11, 2021
Software

All your messages. One inbox.

Inbox zero for messages. All in one chat app with Superhuman-inspired archive + snooze system. Reverse engineered iMessage, WhatsApp, Messenger, and Twitter. Built something that people wanted.

Live product
Lots of features people loved.
Nov 1, 2020
Hardware
Tesla

Tesla Dojo: The Fastest AI Supercomputer

Landing Image

I was the primary engineering program manager on Dojo, Tesla's AI training supercomputer.

We were in the middle of rewriting the Autopilot software stack to go from feature-based ML to end-to-end, back box neural networks.

We needed a ton of compute to train the networks. So we built Dojo, a custom supercomputer optimized for neural network training.

Some press:

Dojo supercomputer
Nov 1, 2020
Hardware
Tesla

Tesla Thumbdrive

What's the first place you drive to after you buy a Tesla?

Sadly, it's an OfficeMax, to buy a damn thumbdrive to store your dashcam footage (and play games).

Worse, those cheap thumbdrives would often fail in the field. Writing 8 cameras worth of video constantly is a tough job for a thumbdrive.

So, Elon had us do a "fun" 1 month sprint: design the best performing (and best-looking) thumbdrive, and put it in every Tesla we sell.

You might still be able to purchase it here. Not sure. I have like 50 test units in storage.

Some press coverage:

Nov 1, 2020
Hardware
Tesla

Tesla Radar Heater

To prove out Autopilot as the superior self driving technology, we need to show that it can handle all the edge cases that a human driver can. One huge edge case is driving in the snow.

Any sufficiently advanced self-driving software stack is robust enough to handle snowy conditions— being careful with accelerating or braking too fast. But the real issue is hardware: the front bumper of the car getting covered in snow and ice would occlude the front-facing radar, which makes Autopilot unusable.

We tried loads of different solutions— hydrophobic coatings, paint treatmens, etc. We landed on a metallic heating element— kind of like a heated blanket— wrapped in an adhesive that we partnered with 3M to manufacture.

There were two engineering problems we had to solve to get this into production:

  1. Seeing through metal: Metal traces and electromagnetic waves don't mix well. Think of how your phone loses reception in an airplane.
  2. Control: It would be too costly and high-failure to put a temperature sensor under the car bumper. We had to figure out a way to control the temperature of the heating element without a sensor— a nearly open-loop system.

For solving both problems, I received a patent along with a few other engineers.

Patent here.

Some press about the radar heater:

Oct 1, 2020
Hackathons

IvyHacks

The First Ivy League Hackathon

3000+ attendees. 6 Ivy Leagues. Fully virtual.

A hackathon buddy and I cofounded this virtual hackathon because we understood a simple equation:

A bunch of poorly run virtual hackathons from each Ivy League < < one single well-run virtual hackathon from all Ivy Leagues.

See some of the submissions here.

Nov 21, 2019
Writings

Seeking Discomfort

On transportation & human centered design

I told myself I wouldn't tell my kids these stories. And yet, posted today to the immutable internet, here are the stories of how my roommates and I ended up having Thanksgiving dinner at a Chinese takeout restaurant, and how we went without food and shelter in the mountains of rural China — two separate stories, of course.

It was my very first Thanksgiving away from home. Instead of family and my aunt's savory turkey, it was my roommates and a d-hall Thanksgiving meal — I was stuck in Boston for the weekend. But the worst part of that dinner was missing it entirely: we misread the emails and showed up too late. And by the time we made the bold move to have Thanksgiving dinner at the Hong Kong, a formidably fast fast-food takeout joint, I was in an Uber, racing to the suburbs of Boston to pick up my shift as a server for an affluent family's Thanksgiving. I missed Thanksgiving dinner back home, at Harvard, and even at the Kong. My Thanksgiving dinner at 10 p.m. consisted of a handful of breath mints and half a family-sized bag of Funyuns.

We continued to make the best of this odd Thanksgiving. After feeding a floormate's pet fish in their common room, we decided to overstay our visit and watch a movie on their TV. You know it's been an odd Thanksgiving when you fall asleep on the bed of a near-stranger, platonically cuddling with your best friend for warmth as he confides in you about a recent breakup.

There was something wrong with every part of that night. But we found a way to have unforgettable Thanksgiving, and it didn't cost a dime. I wasn't able to put it into words, but I knew I craved for more experiences like that night.

Two years later, in the mountains of rural China, that feeling resurfaced. Allow me to paint the picture.

Six college students, two parts brave and 98 parts dumb, spend a summer night in the Huangshan mountains, with no hotel, hostel, or even a tent prepared in advance. Armed with one meatstick per person and a singular loaf of bread, they find a clearing in the forest and try to sleep on a stone slab by a cliff. As temperatures drop, temperaments rise, and one sleepless traveler nearly runs off the edge of the cliff in hysteria. They move camp and break into an abandoned shack for some shuteye. Two hours later, stomachs rumbling, eyes sunken, they summit the mountain to witness a beautiful East Asian sunrise. Of course, they promptly pay the price with a hungry 14 kilometer trek back down to civilization.

I'm often in disbelief that I am of the (un)lucky six who experienced Huangshan this way. I used to be a huge control freak: it was once anxiety-inducing to not know where I was, where I was headed, and why I was headed there. It took experiences like my freshman Thanksgiving and Huangshan — experiences where my control was heavily limited by the environment — for me to be comfortable giving up control.

Investor and author Tim Ferriss argues that a great way to judge situations of high uncertainty is through fear-setting: playing out precisely what the worst-case scenario looks like. An untraditional Thanksgiving wouldn't destroy my social connection with my family. A night out in the mountains would probably get me sick, but the path to recovery is clear. It was by realizing that the worst case scenario was not so bad that the experience became de-risked, leaving me happy to give up control.

As the YouTube group Yes Theory often discusses, there is one primary question we need to ask more often: What makes you uncomfortable? They contend that most personal growth occurs by deliberately venturing outside your comfort zone. By accepting a loss of control and leaving it to fate, or God, or quantum randomness — whatever name fits you — I found a lot of peace in being unprepared. Before, this statement would have shaken me to my core, but now I'm actively working on honing my ability to take on the grey areas with a new calmness.

Maybe it didn't have to take a near deadly overnight hike in China or even a soiled Thanksgiving to discover the rush in pursuit of seeking discomfort. Still, I take away the value of keeping a high spirit in embracing the ugly uncertainty of life.


Originally posted to The Harvard Crimson.

Nov 7, 2019
Writings

Conducting Through the Chaos

Put on your hard hats. We're talking transportation.

Imagine you are the conductor of the New York Philharmonic. Your musicians switch up their instruments at will, and they come and go throughout the piece. Can you create a beautiful performance out of this cacophony? You're going to need a really big baton.

While no self-respecting musician would entertain this hypothetical, the analog lies at the core of legislators, engineers, and researchers in the transportation industry. Managing the rhythm and meter of how urban residents commute through their city is much like a conductor managing what seems to be a chaotic ensemble. One particularly personal problem in transportation that you and I will soon have to deal with is the move of the School of Engineering and Applied Sciences to the new Allston campus. By fall 2020, many a quadling will be burdened with the following question: How are we going to commute nearly four miles roundtrip between dorms and classes?

We are not the first to think of this difficult and interesting engineering problem. In anticipation of the Allston move, SEAS is creating 570 bike parking spots, a bike sharing hub, an “enhanced” campus shuttle system, and a long-term vision of a commuter rail station. But these won't be enough. I don't know about you, but when it gets below freezing and icy outside — a condition for a good portion of the academic year — I wouldn't be very gung-ho about taking the four-mile trek by bike just to get to class. With bikes out of the picture, shuttles will inevitably get overcrowded: Thanks to Harvard's block schedule, most students with a 9 a.m. class will shoot for the 8:45 a.m. bus, not the 8:30 a.m. And if you can't get a spot on the bus, are you out of luck?

Luckily for us, industry and academia also have a whole suite of proposed solutions. The Massachusetts Institute of Technology developed the Persuasive Electric Vehicle, an eccentric miniature car-bike hybrid that can be hailed from ride-share apps and ridden in the bike lane. On our side of Cambridge, Harvard's Human Powered Vehicle Team and a team of 45 students in ENG-SCI 96: “Engineering Problem Solving and Design Project” have also taken a stab at the problem. And, of course, you can take my personal favorite mode of transport and ride a last-mile electric vehicle, like Boosted's electric skateboard or Harvard student-run mobility company Scoob's electric scooters. Many think the answer to our transportation woes must lie in some precarious balance of scooters, bikes, buses, and trains.

But in my opinion, the solution begins with rethinking our perspective: this isn't just an engineering problem. It's a socio-technical one, where the social challenges are just as important as and deeply intertwined with the engineering challenges.

While the initial solutions seem fairly reasonable, I don't expect SEAS to get it right the first time around. The Allston transportation problem is so hard to solve, let alone define, because we students don't even yet know what is best for us in this new upcoming commuting culture. What is important is that we take a people-centered approach to problem-solving. We must define the success of Allston transportation not solely on whether the bike racks are built and the buses are on time, but largely on whether students are satisfied with how they commute. Only then can we take the right steps forward to converge on the best solution for us.

The engineer in me admittedly gets excited by all the new gadgets we can throw at the problem of transportation more broadly — compact vehicles, high-speed monorails, and even jet suits, to name a few. However, gadgets alone don't solve problems. I've taken the time to get equally informed on and excited by the “why”: why solve transportation?

First, to solve transportation is to create a more sustainable future. When we need to get to another part of town quickly, how do we change the narrative so that we don't immediately reach for the Uber app? What new affordances are we given to design cities when the central facet of commuter culture is no longer the gasoline engine?

Next, to solve transportation is to increase our collective productivity. The opportunity cost of effective human hours lost while stuck in traffic is astounding. If we do it right, many urban dwellers have a shot at reclaiming hours or even days of wasted time back.

Finally, to solve transportation is to change the way we feel about life in a city. Too many of us spend too much time in this limbo-state of dissatisfaction and constant urgency while commuting. I dream of a future where this part of our lives becomes just a little less monotonous, making us just a little happier on the average day.

Transportation is just one area where human-centered design and problem-obsession can give us huge societal return. But the ethos is replicable. As the conductor of your own busy life — a life of relationships, career-building, family, and, of course, transportation — I invite you to think deeply about how you direct your musical composition. How do we optimize our lives to balance efficiency, sustainability, and pleasure?


Originally posted to The Harvard Crimson.

Nov 7, 2019
Writings

The Root of Harvard's Suffering

Harvard is poorly designed.

It's time to address the elephant in the room. While Houses, concentrations, and choice of extracurriculars may divide us at times, there is one common enemy that we all face on a daily basis: Harvard's unintuitive doors.

It's one of those things you don't notice until you start looking for it, and when you see it, it will make you sick. Every minor inconvenience, every ephemeral bout of frustration, adds up to a cumulative suffering. Clearly, this is the true source of undergraduate dissatisfaction with campus life. Don't believe me?

Let's start with LISE Cafe. After pretending to do work in there for a bit, you make your way to the door. You see a handle and pull it. It doesn't give. You look up and pull the other door's handle — maybe one of them is locked? Nope.

You read the sign next to the handle: The glass door reads “PULL” on the other side of the door. You visualize yourself on the other side pulling the door. Therefore, you must push. You push the door.

I shouldn't have to know English to know how to open a door. Not only that, but who pushes a door handle?

Angered at LISE's absurdity, you head back to Leverett House to catch some sleep. It's late — cut through Smith Campus Center. You're greeted by two sets of French doors — four doors in total, and only one badge reader. Which door will it open?

You swipe and hear the right set of doors unlock. You go for the rightmost door. It doesn't budge. You go for the left. It opens.

Flustered yet again, you stop by Quincy House's Stone Hall to cry to a friend living on the fourth floor. You tap your ID at the elevator's badge reader. It stays red. You tap again. Still red. You wait a few seconds, and it returns to green. You tap again. Red.

An impatient sophomore grunts past you, hovers their ID two inches from the reader for a one-Mississippi-two-Mississippi, and up to the fourth floor you go. Your eyes swell with tears by floor three.

“What's wrong with me?” you ask yourself.

Nothing is wrong with you. If you also find yourself frustrated trying to figure out how to open a Harvard door, know that you are not alone, and that it is never your fault. Someone designed that door with you, the end user, in mind. But they failed to focus intently on the experience. If you don't know what to do with these trivial devices as soon as you approach them, and you had to have outside knowledge — English comprehension, a Ph.D. in Electrical Engineering, or even the presence of a Quincy sophomore — that's bad design.

As Don Norman writes in “The Design of Everyday Things,” good design is discoverable. That is, just by looking at the object, you should be able to form a mental model of how it works. LISE violates the mental model that handles are to be grabbed, and grabbing is a pulling motion. Smith offers no cues as to which door will unlock. Quincy requires word-of-mouth knowledge and practice just to operate an elevator.

Harvard often fails to use what engineers call a human-centered approach to problem-solving. In this discipline, people are paramount: Cost, engineering, aesthetics, and even time come secondary. While difficult doors are more whimsical indications of our school's poor design, I invite you to question where else Harvard has failed in putting us first. Was Harvard made with people like you and me in mind, or were we the afterthought? What other doors, physical and otherwise, does Harvard shut off to people like us by design?

Roughly four percent of applicants can enter the doors of Harvard. Of the fortunate, one percent can actually manage to open the doors.

There's little I can do alone, but collectively, we are a force to be reckoned with. We don't have to accept these failures of design. Together, through petitions online, sit-in protests in front of the Smith Campus Center's doors, and a sprinkle of administrative help, we can stunt the root of Harvard students' suffering.


Originally posted to The Harvard Crimson.

Oct 24, 2019
Writings

Essentialism

The case for less is more.

The world is noisy. As students, we are tasked with juggling classes, extracurriculars, friends, family, a crumbling political landscape, and questions of our future — all at the same time. For every item worth our attention, there seem to be ten tantalizing pieces of junk that seek to win over our focus. How do we differentiate the two? And even when we can, how do we pick the right important items to focus our time and energy on?

Communication engineers have found the intuition to solve this dilemma over a hundred years ago without even knowing it. Say you want to load up a YouTube video on WiFi. A router is trying to send your phone electromagnetic waves with all the relevant data. However, there are millions of other electromagnetic waves going through the air at the same time, from cellular networks, to AM and FM radios, to Bluetooth. How do you find your wave?

Introducing the bandpass filter.

Most WiFi communications send signals out at a frequency of 2.4 Gigahertz (2.4 billion cycles per second). FM radio is at around 100 Megahertz, and Bluetooth at 2.45 Gigahertz. You can use a bandpass filter to select a specific frequency and throw away all the other information so that your phone will focus on just the WiFi signal. Only then can your cat video on YouTube play.

I've used the concept of the bandpass filter in many areas of my life. In the fall of my sophomore year, I grinded away at countless summer internship applications online — totaling over 150 submitted applications. Of the 25 employers that even responded to me, I managed to get an interview with six, and an offer from one. But those 150 applications were not top quality — they were good ways to fill up my time and convince myself I was doing something productive. The fact and feeling of progress are entirely separate. It wasn't until my junior year that I changed my strategy by applying the simple rules of the bandpass filter. I reached out to five people, interviewed with four, and signed an offer. Less of the bad, more of the good.

Admittedly, this elegant framework for life often bites me back when I'm overzealous with the cutting. Harvard's acquaintance scene can be summarized in one hollow, empty promise: “Let's grab a meal.” The phrase still makes me cringe today. For a while, I vowed not to even suggest grabbing meals with others because I didn't want to be that Harvard kid meeting with 50 “friends” each for 30 minutes on alternating weeks. I wanted a small group of deep, intimate friendships to make the backbone of my social life. Less of the bad, more of the good, right?

But bandpass filters don't have clean cutoffs. Engineers struggle to design filters that perfectly reject 2.39 GHz, and perfectly accept 2.4 GHz. In reality, some of the noise from the fringes will make it through. Now, I see this as a feature instead of a bug. Nonideal bandpasses allow us to focus our energy on the things that matter most, but they also bring in just the right amount of excess to allow us to see something new. I've made some of my closest friends from a thirty minute lunch. And I'm willing to take in some noise if it means a shot at something new.

The rules of the bandpass are seen most clearly in my dorm room. My walls are bare, my windowsills empty, and my desk holds only what I'm working on at the time. My bed intentionally rests at the far end of the nook, blocking me from viewing the work on my desk. Physical spaces are manifestations of our inner state of being, but I've discovered that this is a two-way function: physical spaces also have the power to change our state. For that, I opt for the bandpass in my room. I opt for less is more.

A life of essentialism — or minimalism, or intentionalism, call it what you want — is not a life of deprivation, but rather a life of intentionality. As writer Joshua Fields Millburn puts it in the article “Essentials, Nonessentials, and Junk,” “The key, then, is to continue to question the things we bring into our lives, and to question the things we hold onto, because the stuff that adds value today might be tomorrow's junk.” By navigating my time at Harvard through the framework of the bandpass, I've found more joy in doing and thinking less, and focusing more on what matters.


Originally posted to The Harvard Crimson.

Oct 10, 2019
Writings
Venture

Faux Trophies

My first time getting screwed by a VC.

Let me tell you about the time I got professionally hustled.

It's fall 2017 — my first semester at Harvard. Having never touched the world of tech entrepreneurship, and seeing the startup scene at the Harvard Innovation Labs, the School of Engineering and Applied Sciences, and MassChallenge, I was itching to try my hand at running a venture.

Enter a new inter-Ivy League honors society — let's call it AVC for “Actually a Venture Capitalist” — running a huge New England pitch competition where the top three teams win $10,000 each. I had never seen more than a few hundred dollars at a time — to be able to spend $10,000 working on a project of my own choosing seemed too good to be true. I rounded up two equally naive friends of mine and we got to work.

We spent nearly three months tirelessly preparing the engineering and business models for this competition. With a combined total of three weeks of technical expertise under our belts (i.e., one week of an introductory computer science class), we sought the help of engineers at the Active Learning Labs and created a mechanical prototype. And with not even a single day in a macro or microeconomics lecture), we cold-called dozens of business development gurus on LinkedIn for advice. We quit comps and stunted new relationships to put time into seeing our venture succeed. After many sacrifices and weeks of work later, we had our product.

We called it Kumo Drones. Kumo, which means “spider” in Japanese, aimed to create a highly flexible peer-to-peer drone delivery network in densely clustered college towns for academic researchers to exchange samples and specimens across campuses. We envisioned a future where increased collaboration in research generated faster and more impactful discoveries to help advance humanity.

We submitted our slide deck and sent over all our engineering documents. Kumo Drones was ready for its first win.

The huge competition was not as huge as we had expected it to be. Competing for the top three spots, only three teams showed up: A team of Ph.D. students four years into creating their product, another team of business school students engineering a new cryptocurrency coin, and us, three baby-faced amateurs fresh out of high school.

Surprise, surprise: We placed top three. We sat on a coffee table with our slide deck pulled up on my laptop and pitched our venture. After some half-hearted nods and grunts, we got a pat on the back and a photo with all three teams huddled around one certificate that I could have gotten from a Dave and Buster's birthday party. That was it. A few days later, AVC's emails bounced, its website went offline, and it had all of our intellectual property.

I was mad. After putting our blood, sweat, and dropped problem sets on the line, we came home with nothing. Saying yes to this venture in effect made me so say no to hundreds of other opportunities. I felt like I wasted my freshman fall.

So I told the story in a different way. I told myself that I placed top three in my first ever pitch competition against teams of MBAs and Ph.D.'s — technically true, but truthful omission in reality. I kept this version of the story to convince myself that I was a natural in the entrepreneurial world. Deep down, though, I knew this was a lie.

It took me a while to finally swallow my pride. Today, I'm okay with saying my first “win” in entrepreneurship was not as impressive as I'd like it to be. I'm okay with having been that wide-eyed freshman, optimistic and unassuming of others. And I'm okay with asserting that while I've learned a lot in this journey, I've only scratched the surface.

The stories we tell ourselves dictate our truths. The power of controlling the narrative is a power I still struggle to deal with today. Before, I called myself an early success ready for more wins. But that story doesn't take into account the setbacks and struggles that come along the way. So I've revised it: I'm just a guy who's been lucky enough to mess up so many times that the right path becomes slightly clearer by the day.

Two years later, I'm thankful to have been hustled my freshman fall. Though we didn't make it out with $10,000 or even a certificate of participation, I take from that experience something greater. At the end of my time at Harvard, I don't want to leave with just a list of accomplishments — prizes awarded, funds raised, titles earned. I want to leave with a killer toolbox. I left the competition with a faux trophy. Two years later, I have better tools, and the lessons I learned that semester endure in everything I create.

Consider this an open letter to AVC, from the Kumo Drones team: Thank you for leading us down an unanticipated path. We're better because of it.


Originally posted to The Harvard Crimson.

Sep 12, 2019
Writings

Venturing Past Harvard Yard

Leaving Harvard before it was cool.

Leaving the gates of Harvard College was the best decision I've made in my college career.

When tourists and overzealous high school seniors ask what your favorite part of Harvard is, the “right” answer is that it's the people who make Harvard worthwhile. But what does that translate to in practice? There are people, good and bad, at literally every institution, by definition. Why is Harvard worth fighting for over many other colleges? Three years into my time here, I'm happy to report that Harvard isn't terribly special: It's everything that the College enables that makes Harvard the place to be as an undergraduate.

I am thankful to have learned this lesson early on: Eight weeks into freshman year, after I had just joined the organizing team of HackHarvardCollege, an annual hackathon, a group of 10 Business School students approached me and my co-organizers with a pitch. They wanted our expertise to create an event like no other: a social impact hackathon held at the Vatican City with the support of the Catholic Church.

This was quite an oxymoron to me: Hackathons — events of innovation and rapid iteration — are the antithesis of the Vatican City — a theocratic bureaucracy known for being anything but innovative and iterative. The expenses, logistics, language barriers, and newness of it all made it a recipe for disaster. We foresaw months of planning, weeks of pursuing what would become dead ends, and days of insurmountable stress to bring life to this oxymoron. And yet, we saw what it could be: a future in which the Catholic Church, an organization with a reach on the order of billions, champions new technologies as a vehicle for social change. There was so much for the world to gain, and a lot for us college students to lose.

So, of course, we bit.

We somehow did it. We found 120 of the brightest minds across every continent and flew them all out to the Vatican in less than six months' time. While it was great proving to the world that faith-based institutions like the Church have both the ability and the obligation to lead global humanitarian efforts, I learned another equally profound lesson: Business School students are rockstars. With their marketing and business experience and our hackathon logistics experience, we made the perfect team. The larger Harvard community, where ideas and skills merge to form a collective power, is strongest together.

In navigating Harvard's cutthroat pre-professional scene, I also found that the most rewarding opportunities can from the most unlikely of places. In the lull of the sophomore-slump life, I found myself scrolling mindlessly through Facebook videos until I stumbled upon a video of a man flying a jet suit.

Flying humans — I felt another oxymoron brewing. It was time for another rodeo: this time in a term-time engineering co-op at Gravity Industries, a U.K.-based startup building jet suits. Five micro-jet engines (two on each arm, one on the back), a bladder of A1 kerosene jet fuel, and some fancy electronics give you the experience of untethered human flight. I didn't earn any course credit, money, or Iron Man fantasies fulfilled during my time at Gravity — but every minute at this company focused both on future modes of transportation and inspiring the youth through STEM education was worth it.

Aside from the weird flex, the stories of VHacks and Gravity Industries are an invitation to question your time at Harvard: Where can you take a risk? Where can you do something stupid and crazy where failure is likely the only option? What can happen if you succeed? And, most importantly, how bad will it be, really, if you fail?

As I see it, Harvard is the safety net that should enable us to reach far past the gates of Harvard Yard. With guaranteed food and shelter, access to jobs and internships, and that sweet, sweet grade inflation, the College is comfortable. But being comfortable is dangerous — it can either make us very risk-tolerant, or very risk-averse. I implore you to choose the former.

In pursuing my off-beaten path at — or more precisely, outside of — Harvard, it is because of this safety net that I've felt confident to venture into new lands (and air). The College is just the beginning: As remarkable as it may appear, a look outside the hobbit-hole reveals something greater.


Originally posted to The Harvard Crimson.

Jan 9, 2019
Hardware

Phone as a Key for Electric Skateboards

Demonstration here.

The Problem

I own an electric skateboard made by Boosted, and it is my main form of transportation in Boston. I worry too often that someone might pick it up and make a run with it (they ain't cheap), so I needed some way to curb my anxiety.

Design Considerations

I need something that is:

  • Human-focused: minimizing friction for the user. It should be as easy as walking away from the board to arm it.
  • Energy efficient: ideally, this should last me several weeks before having to recharge.
  • Cheap: #collegestudent.

Proposed Solution

  • Inspired by Tesla's keyfob technology, I'll use wireless signals from a phone to determine distance to the board, which will control the system state with no human intervention.
  • We'll use off-the-shelf components and focus on minimizing quiescent current draw.
  • We'll "borrow" as much hardware as we can from Harvard's maker labs.

Building

MBOM

  1. Wireless receiver
    • WiFi was my initial gut reaction, but the user would not be able to use other WiFi networks while using the product. Out!
    • Cellular signals are nice, and we can always measure pings, but confirming a device ping is my device and not another, say, T-Mobile device pinging on the same band would prove problematic.
    • Bluetooth seems the way to go. It can be connected to my phone as a peripheral, and should give me a pretty good idea for distance when looking at signal strength. This BLE UART from Adafruit ought'a do it (Fig. 1).
  2. Movement/SLAM sensor
    • An accelerometer is a good way to go. Since I don't have one at the moment, for now I'll go for a tilt ball switch I had laying around (Fig. 2).
  3. Speaker/alarm
    • Cheapest thing I could find was a Piezo buzzer. They're small but loud-- and dirt cheap.
  4. Microprocessor
    • I'll start off with an Arduino Uno for prototyping, but for the final product I'll elect to choose either an Arduino Trinket or a Micro.

Calibration

I need to get the signal strength of the connection between my phone and this receiver. After soldering the header pins, I was able to use a SoftwareSerial connection to request via UART the received signal strength indicator (RSSI) value via command "AT+BLEGETRSSI" from my TX to the receiver's RX.

Signal Processing

First, the motion sensor to sense whether the board is moving:

What, this component isn't interesting enough? Fine-- I'll implement it with as an interrupt service routine. Woo-hoo.

Understanding Piezos

A piezo buzzer works by exciting a ceramic material particular distances at different voltages. You can use pulse-width-modulation to control the frequency to produce different sounds. Here's how I designed the sound system:

  • playTone: plays a tone via changing the duty cycle with tone value for duration seconds.
void playTone(int tone, int duration) {
  // Excites buzzer up with tone duty cycle for duration time. 
  for (long i = 0; i < duration * 1000L; i += tone * 2) {
    digitalWrite(SPEAKER_PIN, HIGH);
    delayMicroseconds(tone);
    digitalWrite(SPEAKER_PIN, LOW);
    delayMicroseconds(tone);
  }
}
  • playNote: maps tones to standard notes a through g to make it easier to play the right notes.
void playNote(char note, int duration) {
  char names[] = { 'c', 'd', 'e', 'f', 'g', 'a', 'b', 'C' };
  int tones[] = { 1915, 1700, 1519, 1432, 1275, 1136, 1014, 956 };

  // play the tone corresponding to the note name
  for (int i = 0; i < 8; i++) {
    if (names[i] == note) {
      playTone(tones[i], duration);
    }
  }
}
  • playSound: Simplifies music design, grabbing the notes, tempo, etc. and toggling a quiet variable to ensure two tunes aren't played at the same time.
void playSound(String notes, int beats[], int tempo, int numNotes){
    quiet = false;
    for (int i = 0; i < numNotes; i++) {
        if (notes[i] == ' ') {
            delay(beats[i] * tempo); // rest
        }
        else {
            playNote(notes[i], beats[i] * tempo);
        }
    
        // pause between notes
        delay(tempo / 2); 
    }
    quiet = true;
}
  • handleSpeaker: Receives high-level audio command and puts that command in the queue for speaker to receive music written in this function
void handleSpeaker(String state){
    if (state == "armed_alert"){
        numNotes = 15; // the number of notes
        notes = "ccggaagffeeddc"; // a space represents a rest
        int ourBeats[]= { 1, 1, 1, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 2};
        tempo = 150;    
        playSound(notes, ourBeats, tempo, numNotes);
    }

    if (state == "arm_toggle_alert"){
        numNotes = 5; // the number of notes
        notes = "bag"; // a space represents a rest
        int ourBeats[]= { 2, 2, 2, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 2, 4 };
        tempo = 150;    
        playSound(notes, ourBeats, tempo, numNotes);
    }

    if (state == "intruder_alert"){
        numNotes = 1; // the number of notes
        notes = "g"; // a space represents a rest
        int ourBeats[]= { 2, 2, 2, 1, 1, 1, 2, 1, 1, 1, 1, 1, 1, 2, 4 };
        tempo = 100;    
        playSound(notes, ourBeats, tempo, numNotes);
    }
}

Controller Design

I chose to start off with Arduino Uno because it gives us sufficient memory and clock speed to not worry about resource constraints. I implemented the code as a finite state machine. I was able to get everything hooked up and get the FSM running before finishing the bluetooth module work. I simulated the input of the system with active low pushbuttons for lock and unlock. That's uninteresting and not elegant. Here is the final truth table:

System truth table. We consider state A case 1, state B case 2, etc. in our source code.

Source code available for the FSM implementation coming soon on GitHub!

Demonstration here.

Jan 1, 2019
Hackathons

HackHarvard Gloabl

An experimental hybrid hackathon pre-COVID. Partnered with Alibaba.

Little news feature here.

Oct 1, 2018
Hackathons

HackHarvard

The First Ivy League Hackathon

Did a little bit of everything for many years. This is where I cut my teeth during undergrad.

Little news feature here.

Mar 1, 2018
Venture
Hackathons

VHacks

I was on the founding team of undergrads & MBAs at Harvard/MIT that made the first-ever hackathon sponsored by and inside the Vatican City, with the goal of kickstarting the Catholic Church’s social enterprise ventures arm. Some stuff I did/helped on in my kitchen sink role:

  • Raised $XXXk from Google, Microsoft, Airbnb, Salesforce, and others (not including LPs).
  • Managed rapid development schedule of vhacks.org with an external design firm.
  • Led a group of Fortune 500 C-level executives at tech companies and Vatican government leaders in assessing pitches and awarding over $XXXk in funding.
  • Designed the end-to-end in-person experience for our 120 technologist participants, 250+ press contacts, and higher-ups in the Vatican.

Some press:

Jun 1, 2017
Hardware

Pomodoro, Analog Version

An analogic circuit-based Pomodoro system that helps you focus on your work.

First analog design project from my first year at Harvard. This electrical engineering class was the hardest topic I’d ever learned, which convinced me it was the only major worth studying.

Read at your own risk here

Jan 7, 2016
High School

Poetry Out Loud

Used to perform poetry at the city, county, and state level.

Here's one of my performances— viewer be warned, it's a bit cringe.

Jun 1, 2015
High School

Project Speak-Up

Co-founded Project Speak-Up, a non-profit dedicated to solving public speaking inaccessibility in K-12 Los Angeles public schools. Taught speech classes for two years at an enrichment academy, and hosted online speaking tutorials for 10k+ users worldwide.

Archive of our old website

The students from one of the classes we taught.

An archive of some of our copy

Our mission is to foster public speaking skills in our generation's youth that will last their entire lifetime. Project Speak Up hosts speech workshops in order to not only develop student rhetoric, but to also bolster their self-confidence. Speaking and effectively conveying ideas are essential skills for success in today's society. We want to expose kids to speech early because it will greatly assist our future leaders in both interview and presentation scenarios. We want to promote confidence to aid kids in many aspects of life, from building interpersonal relationships to solidifying one's identity.

Video Series

Jun 1, 2015
Writings
High School

Propaganda Design

JAFREEDOM

What started as a student body election turned into a revelation into my future.

In May of 2015, I saw my semester and a half of student body experience as “strong” enough to run for the Junior Senator position. After all, thirty weeks of Associated Student Body was enough to represent 1,000 people, right?

Wrong.

I lost by a landslide. The prior Freshman and Sophomore Senator secured her third term in office. Looking back at it, I was foolish to have been demotivated by this loss.

It actually motivated me to work harder. As a Junior Delegate in ASB, I put my best foot forward as the MEND Canned Food Drive and Spirit Week Chairman. Through a week of school spirit and a month of helping the less fortunate locals for the holidays, I developed a reputation as a positive member of my campus. I saw that leadership is not defined by the position itself, but what you do in that position.

Come the end of junior year, I decided to run for the ASB School Treasurer. Why can we afford new Chromebooks for 4600 students, but the band can’t get cleaning supplies for our instruments? Why did [insert team name here]'s budget decrease this year? How can teachers get a six-figure salary, but the school lunches are barely consumable? I sought to answer these puzzling accusations to satisfy my craving for understanding. But I also wanted others to know. A mentality to leadership that I fought to bring forth was transparency: the answers and information I would gain as a leader should be readily accessible to any of the 4600 students I represent that ask.

With a clear goal in mind, I revamped my publicity. Through reading about Hitler’s effective use of bold colors and striking patterns as propaganda. I sought to employ a similar technique. As cynical as it sounds, if millions knew about Hitler during his campaign for Vice Chancellor, I could surely get my name out there for a high school. I worked closely with my summer drawing instructor to get the shadowing of my face just right, the chiseled jaw acting as a symbol of confidence and strength. We finalized JAFREEDOM:

This surge of propaganda sent in ten posters, 900 backpack tags, and 1000 flyers quickly got people talking.

“Is that the Communist running for office?” “Those posters yell ‘Stalin’ in my face. I like it.

I quickly learned my first facet of politics: [almost] all publicity is good publicity. In a sea of hastily-made, handwritten posters with small-fonted and lengthy slogans, my posters served as a stark contrast. It got people talking. You see, when you have two weeks to convince over two thousand people that they should vote for you, publicity to the masses is the only way to go. This led to some introspective thinking a day after my campaign team was working at full blast: Do I want people to vote for me because of a poster? No. I wanted genuine votes to so that people genuinely believed in my mission for transparency in representation. While spending every day at lunch and nutrition talking to my peers about my campaign yielded some strong supporters, that wasn’t enough. Votes aren’t counted on a spectrum, so the quality of votes would have to be sacrificed for quantity if I wanted to win-- or so I thought.

The answer to a genuine AND winning campaign came in the QR. Behind every flyer and backpack tag came a giant square of bar codes with “SCAN ME WITH SNAPCHAT” in bold on top. I utilized something that most highschoolers have, Snapchat, to allow the thousands that I couldn’t talk to individually to get to know more about my campaign. Taking a picture of that square would take them directly to the Official Jafreedom website. This website created a foundation for people interested in the electoral process to educate themselves about not only what I stand for, but also how to vote in a complex online system. With over 1800 student views, I upheld my standards of a morally sound election.

With the arts of propaganda, the politics of publicity, and the spread of my message through the internet, I was elected as the ASB Treasurer. The days, weeks, and months after would create a devotion to the students-- a devotion that will last far past my two semester term.

Jan 7, 2012
Shenanigans

HURPS (Harvard Undergraduate Rock Paper Scissors)

The world's first collegiate rock paper scissors team.

Our mission statement: to spread HURPS far and wide in the Greater Boston Area.

We got approved by Harvard to run this team.

We planned to hold a rock paper scissors tournament in Boston, with Red Bull as the title sponsor. Sadly, that never happened because it was planned for April 2020.

To this day, I only play one rock paper scissors game per year as instructed by our Chief Medical Officer (my college roommate who studied bioengineering).

I am still undefeated.

Jan 7, 2012
High School
Software

Magic Mirror

MagicMirror² is an open source modular smart mirror platform. With a growing list of installable modules, the MagicMirror² allows you to convert your hallway or bathroom mirror into your personal assistant.

I contributed to it when I was ~13 years old. Was one of the first software projects I ever worked on.

Github