Everest and Frankie
I built this game with my toddler Everest and our dog Frankie in mind. Everest loves throwing balls for Frankie to chase - it's a simple yet fun game! I used JavaScript and Canvas, with some help from Replit Agent and Claude Sonnet.
This website was created in Windsurf with Cascade AI - we're living in the future folks! 🚀
About
I'm a PhD Economist and ML Engineer specializing in building robust, high-impact systems that bridge economic theory and machine learning. At Instacart, I architect marketplace optimization systems that handle billions in GMV - combining economic principles with cutting-edge ML. My deep background in economic theory and empirical methods, coupled with hands-on ML engineering expertise, allows me to build systems that don't just work technically, but create lasting economic value.
Get in Touch
📧 j•n•th•n•u [at] g•••l [dot] c•m
📱 5•• [dot] ••• [dot] 5••9
ML & AI
LLMs, Reinforcement Learning, Causal Inference (my favorite!), Deep Learning. Built stuff with Claude, GPT-4, and Anthropic's API.
Data & Analytics
DBT for data transformations, SQL wizardry (ask me about my window functions!), ETL pipelines that actually work at 3 AM.
Engineering
Python, Java, Scala, Ruby, Go. Built everything from marketplace systems to game physics engines.
Web & Games
Built this site with Windsurf AI! Created Everest and Frankie using Replit. JavaScript/Canvas for game physics.
Economics
PhD-level expertise in marketplace design, auction theory, and incentive structures. Specialized in empirical methods and economic modeling using Julia and Python.
Modern Stack
AI-powered development with Windsurf, Replit, GitHub Copilot. Fast prototyping with modern APIs.
Experience
Senior ML Engineer & Economist II, Instacart
2020 - Present | San Francisco
Economic Modeling & ML Systems
Architected the marketplace optimization engine combining economic theory with ML, handling over $30B in GMV annually. Developed novel approaches to multi-sided marketplace equilibrium.
Critical Items System
Created a system that uses LLMs to figure out which items in an order are most important.
Marketing Intelligence
Developed ML systems that measure the actual impact of our marketing efforts.
Research Scientist, Microsoft Research
2013 - 2014 | Boston
Worked with Prof. Susan Athey's team on search auction mechanisms.
Patents & Research
Making Auctions Work Better
Figured out how to handle ads of different sizes in auctions and how to optimize bidding using reinforcement learning.
Smart Marketing
Developed ways to measure the true impact of marketing and target offers to the right customers.
LLMs in Production
Created new ways to use large language models for real business problems.
Recent Thoughts
ML Systems as Economic Mechanisms
I think about how to prioritize items in customer orders. Should we prioritize baby formula or bananas? It's not just about optimizing a single metric - we need to consider the full distribution of possible outcomes.
Building Systems That Just Work
The best ML systems are the ones you never have to wake up at 3 AM to fix! I'm obsessed with building systems that are both powerful and fair.
The Future of ML Engineering
Tech has made it incredibly easy to build and deploy ML systems. But the real challenge isn't in the coding - it's in designing systems that balance multiple competing objectives and remain stable under various conditions.