AI/ML Intern — Shopping Agent
Remote / Palo Alto · Unpaid Internship
This is an unpaid internship. You get real production ownership on a live platform, direct mentorship from the founder, public credit for what you ship, and a reference that reflects it. Strong interns are first in line as roles convert to paid positions as the platform scales.
About AI Fashion Week
AI Fashion Week (aifashionweek.ai) is a social commerce platform where AI-native fashion meets real runways. Designers and models build portfolios, drop collections, and go live; shoppers discover and buy through livestreams, reels, and Jini — our AI shopping stylist. We're a small hands-on team backed by NVIDIA Inception and AWS Activate, headquartered in Palo Alto with our flagship live event in Rome this September.
The Role
Harden Jini, our AI shopping stylist that lives in every chat and livestream on the platform. Jini has real tool-calling (product search, order status, live-event lookup) running in production today — your job is to make it reliable, fast, and honest at scale. Your fixes ship to real users, usually the same week.
What You'll Do
- Debug and fix agent failures: tool-call errors, retrieval misses, hallucinated or empty results, latency spikes
- Improve retrieval quality over our MongoDB product catalog (filters, ranking, zero-result handling)
- Write catch-proof tests: every fix ships with a test that would have caught the bug
- Extend the agent graph (LangGraph JS) toward roadmap tools: availability, cart handoff, proactive livestream product pushes
- Instrument and read production traces — diagnose from logs, never guess
What We're Looking For
- Working knowledge of LLM tool-calling / agent frameworks (LangGraph, LangChain, or similar)
- Solid JavaScript/TypeScript; comfort reading unfamiliar backend code (Node.js)
- Debugging discipline: root-cause first, fix second
- Currently enrolled in or recently completed a CS/ML program (or equivalent self-taught proof)
Nice to Have
- MongoDB aggregation experience; GetStream or chat-infra exposure; eval/testing frameworks for LLM outputs
How We Work
Small team, high ownership, ship-daily culture. Everything we build is verified before it's called done. You'll work directly with the founder. Remote-friendly; Palo Alto optional.
How to Apply
Email us with your name, LinkedIn, and resume attached — the button below prefills the subject for you.