Founding engineer

Computing went from a few mainframes to billions of devices. We believe intelligence will follow the same path.

AI is becoming easier to consume. We’re building a future where it’s just as easy to create: intelligence made for every application, business, device, and individual.

That’s the world we’re building at Overmind. Today, our open-source platform helps teams turn production agent data into specialized models they own, connecting data preparation, evaluation, training, and inference.

We’re looking for a founding engineer with strong backend fundamentals, practical AI experience, and the judgment to turn an unclear problem into a reliable product. You’ll work directly with the founders and help shape what we build, how it works, and how the team ships.

What you’ll own

• Product delivery: Work with users, define the problem, and ship across the backend, API, and React console. Follow through after release and use what you learn to improve the product.
• Reliable workflows: Build dataset, evaluation, and training jobs that handle retries, cancellation, worker failures, and partial results without losing work or duplicating expensive operations.
• Data and evaluation: Turn traces and uploads into reproducible, versioned datasets. Preserve lineage, prevent training/evaluation leakage, and compare model quality, latency, and cost against credible baselines.
• Training and serving: Connect GPU workers, checkpoints, and deployment. Improve progress reporting, failure recovery, model activation, cold starts, and inference performance.
• Developer experience: Keep REST, MCP, SDK/CLI, and console behavior consistent. Make permissions, errors, and long-running operations understandable to people and coding agents.
• Technical direction: Make architecture decisions, simplify systems, and own testing, deployment, monitoring, and production debugging. Build security and project isolation into the design.

What you bring

• Production ownership: You’ve taken substantial software from an ambiguous requirement through launch and operation. You can explain your decisions, the failures, and the results.
• Strong Python: You can design APIs and data models, reason about transactions and concurrency, investigate slow queries, and build dependable background processing.
• Full-stack delivery: You can build and debug React/TypeScript interfaces, manage client/server state, and make complex workflows clear to users.
• Practical AI experience: You’ve built or operated LLM or agent workflows involving tools or multi-step execution, and improved them using evaluations or production evidence.
• Product judgment: You ask good questions, challenge assumptions, communicate tradeoffs, and choose the smallest change that solves the real problem.
• Effective AI tooling: You use coding agents to move faster and remain accountable for understanding, reviewing, testing, and operating what they produce.

We expect depth in backend, platform, or applied AI engineering and the willingness to work across the rest. Shipped work matters more than a particular degree or employer.

Our stack

• Backend: Python, Django/DRF, Celery, PostgreSQL, Redis.
• Console: React, TypeScript, Vite, TanStack Query/Router, Tailwind CSS.
• Data: OpenTelemetry, pandas, PyArrow, DuckDB, S3.
• Training and inference: Modal, PyTorch, Hugging Face, Unsloth, LoRA, vLLM.
• Developer tools: Python SDK/CLI, REST, MCP, Docker Compose, GitHub Actions, pytest, Vitest, uv, Bun.

Be comfortable in Python and TypeScript. You don’t need prior experience with every tool listed here.

Useful extras

• Developer tools, SDKs, MCP servers, observability, or data platforms.
• Fine-tuning, model serving, GPU workloads, or evaluation design.
• Distributed job systems, multi-tenant services, or self-hosted software.
• Early-stage startup experience, open-source contributions, or customer integrations.

Your first 90 days

• 30 days: Ship something real. Get hands-on with the full platform and land your first meaningful improvement.
• 60 days: Own a workflow. Take a feature from customer problem to production and prove it works.
• 90 days: Set the pace. Lead a core area, shape what comes next, and help the team ship faster.

What to expect

• Day to day: Short iterations, direct feedback, changing priorities, and responsibility for getting work into users’ hands.
• Your influence: Help shape the product, architecture, and engineering practices alongside the founding team.

Full time, based in London. Salary: £150,000–£180,000.

How to apply

Send your background and one or two things you’ve shipped. Tell us what you owned, a difficult technical decision, and what happened after launch. Repositories, demos, write-ups, or descriptions of private work all count. Include why Overmind interests you.