Massachusetts Institute of Technology
Master of Engineering
Computer Science, Economics & Data Science
- Expected
- May 2027
- GPA
- 5.0 / 5.0
Education
Massachusetts Institute of Technology
Bachelor of Science
Computer Science, Economics & Data Science
- Graduated
- May 2026
- GPA
- 4.9 / 5.0
Relevant coursework
- Design and Analysis of Algorithms
- Discrete Math for Computer Science
- Machine Learning
- Econometric Data Science
- Applied Probability and Statistics
- Optimization Methods in Business Analytics
- Behavioral Economics
- Linear Algebra
Experience
Jun to Aug 2026 Point72 Data Research Intern, New York Macro signals for discretionary trading, backtested across regimes, plus alt-data forecasts for 26 energy and power companies.
- Built statistical and machine learning models that pull signal out of macroeconomic indicators for discretionary macro trading decisions.
- Researched and backtested macro signals against historical market data to test predictive power and robustness across regimes.
- Used alternative data to forecast key performance indicators for 26 energy and power companies.
Feb to May 2026 Balyasny Asset Management AI Engineer, MIT Generative AI Lab A 7-agent Python pipeline that writes the daily energy market report for 6 BAM markets, then fact-checks itself with RAG.
- Automated daily energy market report generation across 6 BAM Funds markets, replacing manual analyst review.
- Architected a 7-agent Python pipeline with RAG-backed fact-checking to validate market insights against historical data.
- Used LLM orchestration to synthesize, evaluate, and format reports, freeing analysts for higher-value research.
Jun 2024 to now MIT Sports Lab Research Assistant, Cambridge A computer vision "last-touch" detector, built with the NBA, to help referees decide who touched the ball last.
- Developing and training a last-touch detection algorithm in collaboration with the NBA to assist referees in making calls.
- Engineered a NumPy and Pandas pipeline that finds usable video frames in large-scale game datasets for the downstream computer vision model.
Jun to Aug 2025 DraftKings Summer Data Analyst, Boston Machine learning to catch risky user behavior, and an alert tool that made detection sharper with less manual review.
- Applied machine learning and statistical analytics to detect, prevent, and monitor risky user behavior across the platform.
- Developed an Alert Optimization Tool that increased detection accuracy while reducing the manual review burden.
- Balanced risk mitigation against revenue impact in analyses totaling up to $2M.
Apr to Jun 2023 GroundUp Ventures Venture Capital Intern, Tel Aviv Quantitative and qualitative research on portfolio companies, presented in internal firm meetings.
- Researched portfolio companies, quantitatively and qualitatively, and presented findings in internal firm meetings.
Brass Rat Investments
Researched investment ideas for $100,000 multi-sector long-short equity fund at MIT.
Teaching assistant
Linear algebra, optimization methods, industrial organization economics, chemistry, and biology.
Sports editor, MIT newspaper
Wrote and edited every sports story for two years as the section's only member.
Some of My Projects
Trading systems, research tools, and competition entries. One of them runs right here in your browser.
Black-Scholes, by brute force
Code10,000 simulated stock paths price a one-year call option, next to the closed-form Black-Scholes answer.
- Monte Carlo
- $0.00
- Black-Scholes
- $0.00
- Difference
- 0.00%
What am I looking at?
The stock. Every line is one possible future for a stock that starts at $100. Each day it moves up or down at random, about 20% a year in either direction, while drifting up slowly on average.
The option. A call option is a ticket that pays you however much the stock ends above $100 (the dashed line) in one year, and nothing if it ends below. So what is that ticket worth today?
The colors. Orange paths finish above $100, so the ticket pays. Grey paths finish below, so it pays nothing. Red paths crashed below $60 along the way.
Monte Carlo. Simulate 10,000 futures, average what the ticket pays across all of them, and adjust for interest you would have earned in the meantime. That average is the price.
Black-Scholes. The 1973 formula, which later won a Nobel Prize, gives the exact answer for this same model in one line of math. The two prices usually land within a percent or two of each other, which is the point: brute force and the formula agree. Run it again and the simulated price wobbles a little; the formula never moves.
Last-touch detection
A computer vision algorithm, built with the NBA at MIT Sports Lab, that determines which player last touched the ball so referees can make the call. It runs on a NumPy and Pandas pipeline that picks usable frames out of large game video datasets.
Computer vision, Python4,000+ lines, confidentialPrivate repo
Macro trading on Kalshi
I run algorithms to trade Kalshi's prediction markets on macroeconomic indicators.
Prediction markets, Python
Institutional Knowledge
An AI academic advisor for MIT students, built with two classmates. A LangGraph RAG pipeline over 6,300+ courses and the requirements for 84 majors answers questions about prerequisites, degree audits, and double majors.
LangGraph, RAG, GradioFinal project, team of threePrivate repo
MIT Pokerbots
Competed in MIT's poker bot competition, where teams write bots that play a poker variant against each other.
Game theory, competitionPrivate repo
March Madness optimizer
Machine learning picks for the whole tournament bracket.
Python, ML
Colonel Blotto, evolved
A genetic algorithm that breeds troop allocations for the classic game theory war game.
Game theory, Python
MIT iQuHACK
Competed in MIT's quantum computing hackathon, building a quantum computing project over a weekend.
Quantum computing, hackathonPrivate repo
Heart transplant waitlist optimization
Survival analysis of adult heart transplant candidates using national registry data. Kaplan-Meier curves and Cox proportional hazards models track who receives a transplant, who dies waiting, and how patients do afterward, checked with concordance, Brier scores, and calibration plots.
Survival analysis, PythonPrivate repo
koshersuites.org
The website for MIT Kosher Suites: live weather, eruv status, candle-lighting times, and the parsha of the week.
Live site
Research
Some empirical economics papers I wrote in spring 2026. Neither are published anywhere except here.
Thou Shalt Not Crash
How the Church Accidentally Cleared Bergen County's Roads
Bergen County, New Jersey is the last county in the country that still bans most retail sales on Sundays. Using 1.8 million crash records from 2001 to 2018, I find its blue laws cut Sunday crashes by 15.4%, and by 61.5% in Paramus, the county's shopping hub.
The Stephen Curry Effect
How an MVP Season Reshaped the Returns to Three-Point Shooting in the NBA
Across 6,403 player-seasons from 2008 to 2022, high-volume three-point shooters earned about a 21.2% salary premium after 2015. Each extra attempt per game was worth about 4.2%, while accuracy on its own earned nothing extra.
Put a record on
Six songs I love. Click the record to drop the needle.
Warming up the turntable…
Click the record to play
0:000:00
Places I've explored
I love going somewhere new and exploring. Most of the following pictures were taken by me. Spin the globe or pick a place to see where I've been, starting with a skydive over Maine.
Loading the globe…
Take a walk around Athens
Why Athens? I love history, and I'm trying to learn Ancient Greek, so that's where I chose to build my little world.
Walk around Plato's Academy, go to a library with a mysterious diary, and try to win the 'coin flex trophy'. Some of the easter eggs live here :)
BlackjackStadion raceSchool of AthensMinesweeperOutfit shopPlato, in person
Play
Outside the classroom
Favorite song
"Vienna" by Billy Joel, from 1977. It has been my favorite for years.
Learning Ancient Greek
I'm learning Ancient Greek a little at a time, hoping to read Plato in the original someday.
Piano
I've played since I was seven. Lately it's been a lot of Chopin.
Sports writing
For two years I was the sports editor, and the only sports writer, at MIT's student newspaper.
Quantum chess
My favorite chess variant. A piece can split across two squares at once and stay in superposition until a move like a capture forces a measurement. I think superposition is one of the coolest ideas in physics, so here is a knight you can try it on.
Puzzle 1 of 5, Reunion: The knight is 50% on a1, 25% on e1, and 25% on e5. Make it whole again, 100% on d3. (Hint: it can be done in 4 moves.)
Tap the knight to move it, or switch to Split to send it to two squares at once.
How to play
The knight. It moves like a chess knight, in an L: two squares one way, then one square to the side. Tap it and dots mark the squares it can reach.
Move. Tap the knight, then a dotted square. When the knight is in several places at once, each part moves on its own, so tap the part you want to move.
Split. Switch to Split, tap a part of the knight, then tap two empty dotted squares. That part divides in half, one half on each square, so a 50% part becomes two 25% parts.
Merge. Move one part onto a square where another part already is, and the two add together: 25% and 25% make 50%.
Measure. Measuring makes the knight choose one square at random, using its odds. Capturing a pawn forces a measurement too. The Measure button only works in the last puzzle.
Winning. Reach each puzzle's goal in the fewest moves given in its hint. Every move and every split counts as one. Press Reset to start a puzzle over.
Hannah Mode
You said the house password. The site is scarlet and gold now, there is a house banner on the portrait, every easter egg you find earns house points, and clicking anywhere throws a few sparks. To turn it off, type "mischief managed" into the terminal.
Say hi
Email is the best way to reach me. I'm always happy to talk about markets, sports analytics, or research.