Artificial Intelligence Engineer

We are hiring AI Software Engineers to build the agents and pipelines that run accounting operations end to end: document intake, extraction, validation, reconciliation, and posting into enterprise ERP systems.

This is a build-heavy role on a small team. You will own features from problem statement to production, which means shaping the workflow, writing the pipeline, designing the evals that prove it works, and watching it run against real customer documents.

We care as much about how you work as about what you have already built. Our engineers use coding agents every day and ship far more than a team this size normally would. If you already work that way, and you are driven enough to take a vague problem and come back with a working system, you will fit here.

What youll do

• Build and ship AI workflows end to end: document intake, extraction, validation, reconciliation, and write-back to the customer ERP.
• Design the prompts, schemas, and pipelines that turn messy real-world financial documents into structured data a controller can rely on.
• Write evals before you write features. Define the ground truth, measure accuracy against it, and hold the line on regressions.
• Use coding agents as your primary source of leverage. Decompose the work, run agents in parallel, and review what they produce with judgment.
• Take features from a rough problem statement through design, implementation, testing, and production rollout.
• Work across the stack: Python and FastAPI services, React and Next.js interfaces, and the workflow bundles in between.
• Debug production behaviour on real customer data, tracing a wrong number back through extraction, matching, and posting.
• Work directly with the product team and with customers to understand an accounting problem well enough to automate it.
• Feed what you learn back into the platform as shared components, better prompts, and faster deployments.

What were looking for

• 2 to 3 years of experience building and shipping production software.
• Strong programming fundamentals, ideally in Python and TypeScript.
• Hands-on experience building with LLMs: prompting, structured output, retrieval, tool use, agent loops, and measuring whether any of it actually works.
• Daily, fluent use of coding agents. You can explain how you break a task down, what you hand to an agent, and how you verify what comes back.
• Evidence of agency: work you started and finished without being asked, and shipped software you can point to.
• Comfort with incomplete information. You can identify the right next step and drive it to completion.
• A high bar for correctness. In accounting, a plausible answer that is wrong costs more than no answer at all.
• Clear written communication. A lot of this teams thinking happens in writing.
• Appetite for hard problems and long stretches of focused work.

Nice to have

• Experience with document understanding, OCR, or information extraction from PDFs, scans, and spreadsheets.
• Experience building eval harnesses or benchmarks for LLM systems.
• Familiarity with FastAPI, SQLAlchemy, MySQL, Redis, and job queues.
• Familiarity with Next.js and React for internal and customer-facing interfaces.
• Exposure to finance, accounting, or ERP systems such as Oracle Fusion, SAP, NetSuite, or Tally.
• Open-source contributions, side projects, or anything you built because you wanted it to exist.