← back
Small AI Teams with Huge Impact — Vik Paruchuri, Datalab
Original: Small AI Teams with Huge Impact — Vik Paruchuri, Datalab
Takeaway
Small teams of senior generalists with AI leverage and boring tech can ship state-of-the-art models faster than traditional org charts.
Summary
- Datalab: 40k GitHub stars (Marker, Surya), 7-figure ARR, team of 4 (was 3); 5x revenue growth since January with Fortune 500/AI-lab customers.
- Prior company Data Quest went 30→15→7 people in layoffs; productivity rose each time, leading to thesis that headcount ≠ productivity.
- Adopts Jeremy Howard's playbook: <15 senior generalists, fill edges with AI/internal tooling (FastHTML, Monster UI), boring tech (no Kubernetes, server-rendered HTML+HTMX+Alpine).
- Trained Surya OCR 3 (500M params, 90 languages, 99% accuracy, character-level bounding boxes, PDF text grounding) end-to-end with just 2 people handling customers→architecture→data→inference→product.
- Demands maturity, in-person work, aggressive component reuse, modular code that AI can extend (rearchitected Marker for this).
small-teamsgeneralistsdatalab
Original description
We scaled Datalab 5x this year - to 7-figure ARR, with customers that include tier 1 AI labs. We train custom models for document intelligence (OCR, layout), with popular repos surya and marker. I'll talk about a new approach to building AI teams, including lessons I learned from Jeremy Howard, and how we manage building popular repos, scaling revenue, and training models with a tiny team. About Vikas Paruchuri CEO of Datalab Recorded at the AI Engineer World's Fair in San Francisco. Stay up to date on our upcoming events and content by joining our newsletter here: https://www.ai.engineer/newsletter