Jonathan Gu

ML engineer, economist, and builder.

I work on marketplace decision systems at Instacart and build independent products that learn from real use. My background is a mix of production ML, causal inference, forecasting, auctions, and dynamic systems.

Outside work, I'm currently building Bountiful Garden, OpenClawBrain, and Project Pelican.

Current Role

ML Engineer at Instacart working across inventory, growth, and marketplace systems.

Training

PhD Economics, UCLA. BA Statistics & Economics, UC Berkeley.

Current Builds

Neighborhood software, bounded memory for agents, and autonomous trading research.

Selected Projects Instacart Work Resume
At Instacart

Production ML systems for a live marketplace.

My day job is the clearest answer to who I am professionally: I build and ship ML systems that help decide what customers can buy, what offers they see, and how marketplace decisions get made under real operational constraints.

What the work looks like

At Instacart I've worked on inventory intelligence, real-time availability estimation, growth incentive targeting, channel decisioning, ads bidding, and auction systems. The common thread is decision quality under scale, latency, and traceability requirements.

  • Built batch and real-time inventory estimation systems, including thresholding and calibration for marketplace availability.
  • Worked on customer growth and experimentation systems where guardrails, causal interpretation, and operational clarity matter.
  • Built ads bidding and auction components with strong emphasis on reproducible pipelines, feature parity, and observable behavior.

Inventory Intelligence

Availability forecasting, batch estimation, and production decisioning across a large, messy retail surface.

Growth and Experimentation

Incentive targeting, channel choice, and measurable product decisions grounded in causality instead of hand-waving.

Ads and Auctions

Automated bidding and marketplace mechanisms where calibration, incentives, and system behavior have to line up.

Selected Projects

Three current bets outside the day job.

I treat this page as a portfolio hub. Each project should be legible on its own, and the deeper product detail should live where it belongs.

Bountiful Garden neighborhood harvest-sharing interface
Live product Neighborhood software

Bountiful Garden

Bountiful Garden is a neighborhood harvest-sharing network for the produce that usually goes to waste. The product turns backyard abundance into real local exchanges instead of another passive community feed.

This is my clearest founder-style project right now: product design, marketplace mechanics, and the local trust surfaces needed to make a simple exchange actually happen.

Marketplace Community Full-stack product
Visit Bountiful Garden
Important project Agent memory

OpenClawBrain

OpenClawBrain gives OpenClaw a bounded memory layer. It pulls signal from prior work, learns in the background, and serves a small useful slice at runtime.

Install / Upgrade How It Works Proof
Project Pelican autonomous options trading research
Private R&D Autonomous trading

Project Pelican

Pelican is a private research system for autonomous options trading. It combines feature generation, model updates, position sizing, and execution logic into a monitored decision pipeline.

It is still an internal research lane, but it reflects the same interests that show up elsewhere in my work: learning under uncertainty, bounded risk, and systems that can keep operating without constant hand-holding.

Quant finance ML pipelines Autonomous systems
Pelican Overview
Writing

A few public artifacts.

Writing is where I try to make the underlying logic visible: what shipped, what the tradeoffs were, and what actually seems true.

Contact

Best fit conversations.

I'm most interested in applied ML systems, agent infrastructure, product strategy, local-first software, and careful marketplace design.

Reach me directly

Use whichever channel is easiest. Email is best for thoughtful project or collaboration notes.

Email LinkedIn Twitter / X GitHub Resume (PDF)