AI that finds the right answer, and proof that it does.
I help teams ship search, retrieval and domain-tuned LLMs that work for real users, not just in a demo. I've done it for 12 years at Walmart, Pinterest, Roblox and Shopify, most recently training Shopify's commerce foundation models.
Taking on a few select US engagements in retrieval, ranking and LLM training.
From search ranking to foundation models.
I've spent my career on the systems between a person's intent and the result they're looking for. I work across the full ML lifecycle: embedding and keyword retrieval, ranking, LLM pre-training and post-training (DAPT, SFT, DPO), and the evaluations that prove it worked.
I built relevance for Walmart.com and the first Walmart Grocery search engine, was a search-quality tech lead at Pinterest, and led search and recommendations at Roblox. At Vantage Discovery I built an LLM-powered search engine that combined semantic understanding with keyword precision. After Shopify acquired Vantage, I joined the Foundation Models team in Shopify's Search org and built commerce foundation models.
In July 2026 I left Shopify to found InferSense Labs. Now I bring that same depth to a few teams at a time, as their fractional AI lead or advisor.
A narrow focus, on purpose
I only take on niche, technically hard problems in retrieval, ranking, and LLM fine-tuning and pre-training. It's where twelve years of depth matters most. If your problem sits outside these, I'm happy to point you in the right direction during our intro call.
LLM pre- & post-training
Domain-adaptive pre-training (DAPT), supervised fine-tuning (SFT) and preference tuning (DPO), from data curation through to a model you can ship.
Search relevance & retrieval
Hybrid embedding and keyword retrieval, ranking and query understanding for search, RAG and recommendations.
Evals that prove it worked
Offline metrics, LLM-as-judge pipelines and A/B test design, so every model change comes with evidence.
Fractional lead & advisory
A senior ML lead for your team part-time: architecture reviews, roadmap and build-vs-buy decisions, and mentoring your engineers.
Engagements are shaped around what you need: a monthly advisory retainer, an ongoing part-time fractional role, or a scoped project with a clear deliverable. We'll work out which fits on a free intro call.
Select clients, senior attention.
I keep my client list short on purpose. You work directly with me, not a junior team behind me, from the first call to the final handoff. I built search and ML in the Bay Area for twelve years and have worked remotely with distributed teams since 2024, so I work the way modern engineering teams already do.
Live time together
I schedule calls, reviews and pairing sessions during your Pacific Time hours.
Progress around the clock
I keep a schedule that overlaps with Pacific Time and India Standard Time, so work moves forward around the clock.
Clear in writing
Design docs, experiment readouts and weekly summaries, so decisions never depend on who was in the room.
A US company
Engagements are contracted and invoiced through InferSense Labs LLC, a U.S. (Wyoming) company, and paid in USD.
Where I've built
- 2026 – now
Founder · InferSense Labs
- 2025 – 2026
Staff ML Engineer, Foundation Models · Shopify
- 2024 – 2025
Applied ML · Vantage Discovery (acquired by Shopify)
- 2022 – 2024
Principal Engineer, Search & Recs · Roblox
- 2016 – 2022
Search Quality Tech Lead · Pinterest
- 2014 – 2016
Search Relevance Engineer · Walmart Labs
Let's talk about what you're building.
Whether you're fine-tuning a model for your domain, fixing search relevance or trying to prove an AI feature works, I'd be glad to hear what you're working on. You'll talk with me directly, not a sales team. I take on a limited number of engagements, so reach out early.