Skip to content

We opened our Constanța studio. See the photos

LABS

Team

A small team, growing on purpose.

A senior founder who ships, two roles we're interviewing for, and an AI stack that does the heavy lifting. We've built with AI since before the LLM era, and it shows in how fast we move.

The team

Senior engineers who talk to you directly and write the code themselves.

  • David Marin, founder of Basetool Labs

    David Marin

    Founder & engineer

    Founds and builds Basetool Labs. Eight years shipping product at YC-backed startups, from frontend architecture to AI features. Talks to clients directly and writes the code.

We're growing

We're hiring senior people who ship. If that's you, say hello. We're interviewing now.

  • Senior Product Engineer

    Interviewing

    React and Next.js, an AI-native workflow, ships product end to end, from wireframes to production.

    Get in touch
  • Founding Designer

    Interviewing

    Owns product and brand design across our own work and our clients'. Senior, hands-on, opinionated.

    Get in touch

Our AI stack

We've worked closely with AI since before the LLM era: a published research paper, then years building in the space. This is the infrastructure that lets a small team ship like a big one, and run models ourselves for IP-sensitive work. (Yes, they're tools. That's the leverage.)

  • Conductor.build

    Agent orchestration

    Our cockpit for running a fleet of coding agents in parallel. The conductor to their orchestra.

  • Claude Code

    Pair engineer · always on

    Writes, refactors, and reviews with us in the terminal. Reads the whole codebase before its first coffee.

  • Codex

    Second opinion

    An independent reviewer on every change. We ship nothing it hasn't tried to pick apart.

  • Promptfoo

    Evals, by the numbers

    Graded test runs on every AI feature. Vibes don't ship; scores do.

  • Local models

    Private inference

    Open models on a 128GB Apple-silicon box, so IP-sensitive code and data never leave the building.

  • RunPod

    Dedicated GPUs · 24/7

    Dedicated GPU instances running models around the clock, so capacity is never the bottleneck.

  • Second brain

    Project memory

    A Mac mini that never forgets: Hermes + gbrain index every project, so context is one query away.

  • Curated skills

    Sharpened on real work

    A hand-picked library of agent skills, tuned on real client builds and reused across projects.

Tell us what you're building.

30 minutes on the calendar. No pitch. We listen, we ask questions, we tell you whether we can help.

Pick which categories of cookies you're OK with. You can change this any time from the footer.