Software Engineer, AI

Ambrook•San Francisco, CA

2d•$115,000 - $280,000•Remote

About The Position

Ambrook is building a world-class team to empower farmers, ranchers, and small-business owners using modern AI tools. You’ll shape how we use LLMs and intelligent systems to make messy, high-stakes financial workflows faster, smarter, and more accessible for users across the country. We’re shipping features that lift our customers out of the drudgery of rote bookkeeping and that help them make smart financial decisions, from cashflow management, to enterprise P&L analysis to long term investment planning. Our goal is to supplement our customer’s talents and savvy with the intelligence of a fortune 500 finance team in the form of an AI CFO.

Requirements

You've built and shipped ML/AI-driven features to production.
You're excited to apply language models to real-world, unstructured, often ugly datasets, and turn them into valuable product experiences.
You're pragmatic about AI: you know when to use an LLM, when to write a regular expression, and when to ask the user.
You understand the value of experimentation in a rapidly evolving world.
You have experience working with LLM APIs, prompt engineering, embeddings, vector search, or fine-tuning.

Nice To Haves

You've worked on products involving transactions, finance, document processing, or communications tooling.
Experience with TypeScript and AI tooling — e.g. Vercel AI SDK, Braintrust or similar eval vendors, and provider APIs (Anthropic, OpenAI, Gemini, Jev).

Responsibilities

Development of intelligent systems and tools that use LLMs, other generative AI solutions and heuristics to streamline bookkeeping, financial workflows and deeper applications that truly understand the user, as well as the infrastructure for evaluating those systems.
Teach best practices around ML system design, evaluation, prompt engineering, and scalable AI infrastructure.
Learn how messy real world financial data really is - from scanned PDFs to unformatted CSVs to text-heavy emails - and how to make sense of it using language models.
Improve our internal tooling and infrastructure for classification, tagging, summarization, and user-in-the-loop AI systems.
Build an assistant that powers a critical bookkeeping workflow.
Collaborate with operations, product and engineering to understand customer workflows and where AI would reduce user pain.
Review our current AI infrastructure and propose improvements to our evaluation infrastructure, harness design, shared memory, durability and similar foundational components.
Act on your proposals from month one: scaling our evaluation pipelines, making our harnesses more extensible, and so on.
Implement or improve systems for document extraction and financial transaction tagging, using both heuristics and model-based approaches.
Collaborate closely with design and product to make AI features feel reliable, intuitive, and helpful – not mysterious or brittle.
Own and scale the systems that form the backbone of our vision for an AI CFO.
Contribute to our long-term AI strategy: which tasks to automate, how to measure confidence, and when to keep a human in the loop.
Help build internal tooling for prompt iteration, embedding search, labeling, and retraining.
Share learnings in a post on Ambrook engineering blog or an open-source tool that helps others tackle similar messy real-world data problems.