Jonathan Gu

I build decision systems for real-world uncertainty.

At Instacart, I work on availability prediction and decisioning: the models, thresholds, and serving systems behind what customers see, what shoppers can find, and how messy inventory signals turn into real marketplace decisions.

Outside work, I build a few things of my own: Bountiful Garden for neighborhood harvest sharing, OpenClawBrain for bounded agent memory, and Project Pelican for autonomous trading research.

The throughline is simple: take noisy signals, decide what matters, and keep the system honest once real users touch it.

Current Role

ML Engineer on Inventory Intelligence / Availability at Instacart, owning customer-facing availability systems end to end.

Training

PhD Economics, UCLA. BA Statistics & Economics, UC Berkeley. Still using those causal and decision-system instincts in production ML.

Current Builds

Bountiful Garden, OpenClawBrain, and Project Pelican — three different ways of turning uncertain signals into action.

Selected Projects Instacart Work Resume
At Instacart

ML for what customers see and shoppers can find.

I work on Instacart's Inventory Intelligence / Availability systems. The job is simple to state and hard to do: predict whether an item is really in stock, then turn that signal into storefront decisions under noisy data, freshness constraints, and real operational risk.

What I actually do

This is end-to-end production ML work. I move across problem framing, training data, feature engineering, model training, batch and real-time serving, experiments, rollout, monitoring, debugging, and migration work when old system paths need to be retired.

  • Led the shift toward a real-time LightGBM availability model with shadow-mode validation, rollout checks, and better observability; the re-envisioned batch real-time model improved marketplace quality.
  • Built thresholding and calibration changes that improved good fill rate and made score-to-product decisions easier to control.
  • Integrated new signals such as Kroger balance-on-hand data and Simbe shelf scans, and built forecasting and evaluation surfaces to figure out what was actually true.

Real-time availability

Models and serving paths for product-store availability, freshness, and customer-facing decisions.

Forecasting and calibration

Day-ahead forecasting, threshold tuning, and evaluation that fits the actual decision problem.

Signals and system hygiene

New inventory signals, pipeline migrations, cost cleanup, monitoring, and the unglamorous fixes that keep a live system sane.

Selected Projects

Three projects I keep coming back to.

These are the clearest examples of the problems I like: local coordination, better memory for agents, and systems that keep making decisions on their own.

Bountiful Garden neighborhood harvest-sharing interface
Live product Neighborhood software

Bountiful Garden

Bountiful Garden helps neighbors share backyard produce that would otherwise go to waste. The goal is not another community feed. It is to make a simple local exchange actually happen.

It is the most founder-shaped thing I'm building: product design, trust surfaces, and marketplace mechanics for ordinary neighborhood life.

Marketplace Community Full-stack product
Visit Bountiful Garden
Live product Agent memory

OpenClawBrain

OpenClawBrain is a memory layer for OpenClaw. It helps an agent learn from prior work without stuffing raw history back into the prompt.

The constraint matters as much as the memory: runtime context stays bounded, inspectable, and fast.

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

Project Pelican

Project Pelican is my private research system for autonomous options trading. It combines data, forecasting, risk controls, and execution into one monitored loop.

It is still internal, but it reflects the same recurring interest as the rest of my work: systems that keep making measured decisions when uncertainty is real.

Quant finance ML pipelines Autonomous systems
Pelican Overview
Writing

Selected writing.

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

Contact

Get in touch.

I'm usually interested in applied ML, agent infrastructure, marketplace design, local software, and projects where the system has to make real decisions.

Reach me directly

Use whichever channel is easiest. Email is best for thoughtful notes about projects, collaboration, or roles that actually fit.

Email LinkedIn Twitter / X GitHub Resume (PDF)