Skip to main content

Ambrook

Software Engineer, AI

San Francisco (Remote)$115k–$280kfulltimemidAdded 1 month ago

About this role

Ambrook is seeking an AI Software Engineer to build intelligent systems that modernize financial management for family-run farms and small businesses. You'll develop LLM-powered features for document parsing, transaction categorization, and bookkeeping automation while establishing best practices for AI infrastructure and evaluation.

What you'll do

  • Develop and ship LLM-powered features for financial document processing and transaction management
  • Design evaluation pipelines, prompt management systems, and feedback loops to improve model performance
  • Build hybrid AI and rule-based systems for automated bookkeeping and financial analysis
  • Collaborate with product and design to create reliable, intuitive AI experiences
  • Establish internal tooling for prompt iteration, embedding search, and model retraining
  • Contribute to long-term AI strategy for automation decisions and human-in-the-loop workflows

What they're looking for

  • LLM development and prompt engineering
  • Machine learning and NLP
  • Document extraction and classification
  • Python and TypeScript/React
  • Production ML system design
  • Vector search and embeddings
  • Pragmatic judgment on when to use AI vs. heuristics
  • Product-focused engineering mindset
Apply with Autofill

Opens the application — the Jobs AI extension fills it for you. Set up autofill

Opens the official application on the employer’s site. No login required.

Ambrook

Ambrook builds AI-powered financial software for family-run businesses in agriculture, construction, and other underserved industries, featuring intelligent document parsing, transaction automation, and modern financial infrastructure. The company is hiring full-stack engineers, design engineers, AI specialists, and recruiting talent to scale its product and team.

View all jobs at Ambrook

Likely interview questions

  • Tell us about a time you shipped an LLM or ML-powered feature in production. What was the biggest challenge in moving from experimentation to a real product?
  • How do you decide when to use an LLM versus simpler heuristics or rules for a problem? Can you walk through an example from your work?