Senior Software Engineer

We’re working with a couple of Series A startups building AI that does real work inside businesses. You’ll be building agents that pick up a messy, human process and run it.

The hard part was never the model. The hard part is everything around it: the platform, the data pipelines, the integrations, the connectors, the observability, the tooling that turns “the LLM can sort of do this” into “this runs in production and the customer trusts it with real money and real decisions.” That’s the job.

What you’d actually be doing

• Designing and building backend systems in Python (FastAPI), from the API contract down to the database schema and the infra it runs on.
• Building connectors and pipelines that plug the product into whatever the customer already runs: their tools, their data, their systems, however messy.
• Writing the internal libraries and tooling that make everyone else on the team faster.
• Owning problems end to end. Nobody hands you a ticket. You get the context, scope the thing, build it, ship it, and own whether it worked.
• Wearing a few hats on a normal day: design a schema in the morning, build the API in the afternoon, debug a deploy before you log off.
• Sitting in product conversations, watching how customers actually use what you built, and having a real say in what gets built next.

This is greenfield. You’re setting the patterns the team builds on for years, not maintaining someone else’s.

The kind of place this is

• Series A, backed by tier-1 VCs. The funds behind the AI companies you already know.
• Small, deeply technical teams. Engineers, AI people, a high bar, not much process.
• Founders who’ve done genuinely hard technical work before joining (think ex-Palantir, ex-serious-infra).
• Based in London, some also in New York. Open to remote for the right person in EMEA.

What good looks like

• 3+ years writing backend software you’re actually proud of. A CS degree is nice, not required.
• You write Python and/or TypeScript to a high standard. Strong typing feels like a tool, not a tax.
• You’ve built libraries, SDKs, or internal tooling that made other engineers’ lives easier, and you enjoyed it.
• You get a small, specific satisfaction out of a clean data model.
• You’re good with ambiguity. A vague problem is something you like turning into a scoped one, not something you wait for someone else to define.
• You’ve worked somewhere small and fast, or it’s obvious you want to.

Don’t apply if

• You want tickets handed to you and a product manager to tell you what “done” means.
• You want to spend your days training models. This is the platform around the model, not the model itself. If building agents is the actual dream, say so and I’ll point you at the AI-engineer roles instead.

#J-18808-Ljbffr…