Richard Wang
01 · Introduction

Richard Wang

Machine learning engineer and founder based in San Francisco. Building tools for software teams and writing in the margins.

San Francisco Applied AI Research & writing
02 · Work

Co-founder & CEO of Clad Labs AI

Clad Labs builds AI infrastructure for software engineering — code retrieval at repository scale, review agents calibrated to production standards, and the evaluation stack that keeps both honest.

Y Combinator · F25 Seed stage Hiring
Visit Clad Labs
03 · Focus

Context for coding agents

The core of my current work: how to retrieve the right slice of a codebase for a model, how to measure whether the answer is any good, and how to make both reliable under real load.

Retrieval Evaluation Agents Infrastructure
05 · In progress

Coming soon

Pieces currently in draft.

Coming soon Koopman operators — a practical intro for time-series ML DMD + Koopman embeddings against chaotic fab-line data.
Coming soon Injecting LLM ads — lessons from YC Vector-DB design, prompt hardening, auction-less ad retrieval.
01 · About

Richard Wang

ML engineer, researcher, jazz guitarist, basketball enthusiast. Co-founder & CEO of Clad Labs AI — YC F25. Writing on applied AI, statistical models, building in public, and whatever else comes to mind.

LinkedIn Twitter GitHub Clad Labs YC F25
02 · Essays

Selected writing

Working notes on AI infrastructure, the research that feeds it, and a couple of things about neither.

    The archive Browse the full index Search, tag filters, essays and notes side by side.
    03 · Media

    Appearances & talks

    Videos, podcasts, and live-stream segments — each one gets a native viewer with chapters and a direct link to the source.

      04 · Notes & updates

      Life, in margins

      Shorter pieces — company updates, talks, and excerpts that don't need the essay treatment.

        All notes & updates The diary view, in the archive.
        05 · In the lab

        Coming soon

        A few longer pieces I'm still drafting.

        06 · FAQ

        Frequently asked

        What tech stack runs this site?

        Cloudflare Pages + Tailwind CSS + Vite + Three.js (for the Klein bottle scroll journey) + a Cloudflare Worker that proxies Google Gemini for the chat widget.

        How does the Klein-bottle scroll thing work?

        A WebGL canvas renders Hermann Karcher's twisting Klein bottle using its parametric equations. Your scroll position maps to a rotation of the surface; the bottle sits on a static lean-back pitch so the ring's hole never stares the camera down. Each scroll gesture advances exactly one chapter — click any dot on the right rail to jump.

        Can I reuse your code?

        Nope, unless otherwise specified. Some of it's MIT.