Career

Where I've worked and what I've built.

  • AI Engineer
    Zip ↗·San Francisco, CA
    Feb 2025 – Present

    Building applied AI systems for enterprise workflows, focused on document understanding, extraction, evaluation infrastructure, and reliable context generation from unstructured data. Leading work across parsing, retrieval, model routing, and AI quality systems. Unofficial title: document processing princess.

    • Document understanding systems
    • Evaluation infrastructure for LLM workflows
    • Context generation from enterprise data
  • Cofounder & CTOYC W24
    Ecliptor·San Francisco, CA
    October 2023 – January 2025

    Cofounded a YC W24 startup that evolved across several enterprise AI problem spaces, beginning with an AI SRE platform before pivoting into sales tooling and eventually document processing for financial services. Led engineering and ML systems development across infrastructure, product, and customer deployments.

    • Y Combinator W24
    • AI SRE + enterprise AI tooling
    • Document processing for financial services
  • Machine Learning Engineer, RecSys
    Twitch ↗·San Francisco, CA
    March 2022 – June 2023

    Worked on recommendation systems and personalization infrastructure at large scale, focusing on ranking systems, user intent, and content discovery. Developed an interest in how algorithms shape behavior and information access online.

    • Recommendation systems
    • Ranking + personalization
    • Large-scale ML infrastructure

Problems I care about

Things I think about a lot.

▶Making AI systems reliable in the real world
The hardest part of AI isn’t model capability, it’s making systems behave predictably under real world conditions. I care about the gap between offline evals and production reality. I want to build systems that actually earn trust over time.
▶Building context layers for AI systems
AI systems are getting better at reasoning, but they still lack a coherent representation of the world they operate in. I care about turning unstructured data (documents, workflows, notes, history) into structured, queryable context that agents can actually use to act on.
▶Diffusing AI through the real economy
Powerful models alone are not enough to transform industries. I’m interested in how AI actually diffuses through the economy beyond pilots. I care about what it takes to make these systems usable inside real workflows and organizations.
▶Change management for AI adoption
Models are becoming capable faster than organizations can adapt around them. I care about the systems, workflows, and institutional changes required for AI to actually reshape how work gets done. The hard part isn’t proving a model works, it’s integrating new capabilities into how organizations actually operate.
▶Expanding access to intelligence
Access to high quality guidance (educational, medical, legal, financial, technical) is still unevenly distributed. I’m interested in a world where intelligence becomes core infrastructure, making expertise and personalized support accessible to far more people, especially through high quality education and individualized learning.

Education

  • B.S.E. Computer Science and Engineering
    University of Michigan, Ann Arbor
    Graduated May 2022
© 2026 Nanki Grewal