Software Engineer

About the role

You will build the features that public-company reporting teams use to draft, review, and file. Cycles are short. Most work goes from problem to production inside a week or two, and you will hear back from the people using it.

The stack is Python and Go services on Kubernetes with a Next.js frontend. You do not need all of it on day one. You do need to be the kind of engineer who reads the code around a change before making it.

Most of what you build sits on top of LLMs: retrieval over filings, agents that draft and check disclosures, and the evals that keep them honest. You will work with the models directly, test their output against source data, and use AI coding tools every day to ship it.

The bar: Our users are controllers and SEC reporting leads working to a filing deadline. “It looked right” does not survive that, so every change ships with evidence it behaves as claimed.

Why this role

What you will do

• Build product features across the Python backend and the Next.js frontend, and pick up the Go services as you go
• Build the retrieval and agent paths behind our reporting features: prompts, tools, evals, and the plumbing that gets the right filing data to the model
• Turn accounting workflows into interfaces a controller can use without training
• Write the tests and evals that check expected behaviour and catch regressions
• Own your features through deploys, production issues, and the customer feedback that follows
• Review your teammates' changes, and have yours read by the engineer who owns the system
• Join the on-call rotation once you know the system, with a senior engineer backing you

What we are looking for

• 2-4 years writing production code, ideally at a product company
• Solid Python or TypeScript, and the willingness to get good at the other. Go when a service needs it
• Product sense. You notice when an interface will confuse a controller, and you say so before it ships
• You read the code around a change before making it, and can say what it might break
• Comfortable with SQL and Postgres, including finding and fixing a slow query
• You use AI coding tools daily and are good with them. You can walk through a change one wrote and how you checked it
• Comfortable with ambiguity. You can turn a short brief into a plan and agree what done means before you start

What adds a bonus

• Shipped something built on an LLM API, and can say honestly where it broke
• React, Next.js, or Tailwind
• Built Docker images or deployed a service to Kubernetes
• Document pipelines, retrieval, or evals
• Interest in finance or accounting. You need the appetite, not the domain
• Side projects, open source, or anything you built because you wanted it to exist

What this is not

• Not a ticket queue. You get handed a problem, not a solution
• Not a role where someone else writes the tests
• Not a rotation through teams. You stay with a product area and maintain what you ship