Goodley

Hiring founding engineers · Remote, UK

Compliance should be easy. Building it isn't.

A small team building AI for regulated firms, with the developer experience we always wanted. Real ownership from day one.

~/goodley · zsh
$ git push origin main
✓ lint0 warnings
✓ typecheckstrict
✓ testreal Postgres
✓ buildonly what changed
✓ migratebefore anything moves
✓ deploylive

01 · the team

Small team. Serious problem. Your name on it.

Join early and own a big part of the product. You ship it, you are on the hook for it, and you have a say in how it is built.

A serious problem

Regulated firms answer to thousands of rules, and their senior people are personally accountable when they get them wrong. Most still run it on spreadsheets.

Domain experts beside you

You build with compliance experts who have done the job, so what a rule means is a question you can ask, not a guess.

Whole problems

Own a feature end to end: schema, API, UI, the AI behind it and the dashboard that tells you it works.

No layers, no ceremony

Work straight with the founder and the firms using the product. Small pull requests, quick review, merge and it ships.

02 · developer experience

The setup you always wanted.

Strict types, fast feedback and nothing to fight. Observability came first: every layer of every request is traced, with logs and metrics beside the trace.

request
controller
use case
repository
query

Strict by default

Strict TypeScript, no any, validation at every edge. The architecture is enforced by the linter, not by memory.

Ship on merge

Merge to main and it is live. Only what changed is rebuilt, migrations run first and a bad rollout rolls itself back.

AI as production software

Typed tools, versioned prompts and every run costed. AI features get the same tests and traces as everything else.

One command to start

A generator scaffolds every new feature with tracing, wiring and tests already in place.

Dev like production

Dev runs the same infrastructure as production, defined in the same code, so nothing surprises you on release.

Agents on the team

We build with AI agents every day, and they follow the same written rules as the rest of us.

TypeScriptBunTurborepoHonoNext.jsReactPostgreSQLpgvectorClaudeAWSTerraformOpenTelemetryDatadog

03 · the work

Agents, retrieval and evals.

We are building agentic workflows for compliance: our own embeddings and retrieval, a harness for agents to work in, and evals to hold it all to account.

embeddings

Embeddings we run

Our own embedding pipeline over regulation and firm documents, refreshed as content and models change.

retrieval

Retrieval you can trust

Retrieval across regulation and a firm's own documents, with every answer traced back to the text it came from.

harness

The agentic harness

The runtime our agents live in: typed tools, approvals, subagents, budgets and a trace for every step. We are building it now, and you would shape it.

agents

Agentic workflows

Agents that plan, call tools and carry real compliance work from start to finish, asking before anything that matters.

evals

Evals at the centre

Where we are heading: every prompt, tool and model change measured against real cases before it ships, so a better model is a decision, not a guess.

team

The best agentic team

We build with agents every day and shape the codebase so they do their best work. You help decide how.

04 · you

You might be one of us if

✓You have shipped production TypeScript and own it after it merges.

✓You open the trace before you guess.

✓You use AI agents as leverage and know when they are wrong.

✓You want a young codebase to shape, with standards worth keeping.

Come build it with us.

Send us something you built and a line on why this problem.

Apply as a founding engineer

info@goodley.ai