# About

Henrik Albihn

I ship production AI for Fortune 500 and Big Four clients at Strange Loop Labs, and I build the infrastructure layer that makes coding agents reliable.

Thesis: Agentic infrastructure is the next platform shift. The substrate around the model — context, search, data access, verification loops — is a 10–100x quality multiplier, and the place where the next decade of venture value accrues.

Receipts

  • Forward-deployed AI for Fortune 500 and Big Four engagements at Strange Loop Labs — shipping alongside ex-AWS, ex-Lyft, ex-Alexa, and ex-Elastic engineers.
  • Owned the size-recommendation engine at True Fit — models serving millions of shoppers daily at the world’s largest retailers: Nike, Macy’s, JCPenney, Target, Walmart, Urban Outfitters, and 30,000+ other brands.
  • Principal AI Scientist on a manufacturing copilot shipped to Meta Quest 3 and Apple Vision Pro.
  • 7 models on Hugging Face — SLERP/TIES merges, GGUF quantizations, Qwen2.5-Sci variants.
  • Youngest IC on a 15-person data science team of Duke and MIT PhDs, ex-Microsoft, ex-eBay, and ex-Spotify ML. Left owning the core product.

Portfolio

Three wedges into the same stack, not three side projects:

  • ticket-rs.io — AI-native issue tracking. Git-backed, dependency-aware, PageRank-prioritized. The context layer.
  • sgrep.sh — Semantic code search. Pure Rust, sub-10ms, Model2Vec embeddings. The search layer.
  • sqlgenie.ai — Natural-language data access across 23+ dialects. The data layer.

What I’m paying attention to

  • Agentic infrastructure — the substrate that makes long-running, tool-using agents cheap, durable, and auditable. Where the moat is.
  • Classical ML for agentic coding — model merging, small-model inference, context compression. The boring parts that beat prompt engineering.
  • Market structure for AI — who captures value, which layers commoditize, where incumbents break. The economics lens applied to an unfinished stack.
  • The feedback loop — we build AI, AI changes how we work, work changes what we build. The cycle tightens. Most analysis underrates the second-order effects.

Trajectory

Economics (BA) → data science → ML science → applied AI → forward-deployed engineering → serial founder. Advising AI teams and early-stage founders on Topmate. Writing here.

Moving toward the capital side of the table — but from the builder’s chair, with working code behind every claim.

Elsewhere