Who You AreYou are an applied AI engineer who ships. You have taken LLM-based systems and agentic workflows from prototype to production, and you know what that requires: evals, observability, cost management, latency tuning, error handling, and iteration.
You have opinions about agent frameworks and architectures, built from what you have shipped. You are also pragmatic enough to pick the right tool for the job rather than force a favorite into every problem. You read new research, try new approaches, and form your own view on what holds up in production.
You work across the stack. You can design a retrieval system, wire up tool calls, build an eval harness, debug a production outage, and explain your choices to a product manager or an executive in the same day.
What You Will OwnYour scope from day one includes:
- Design and delivery of production AI and agentic systems across document intelligence, workflow automation, and copilots
- Architecture decisions for LLM-based systems, including retrieval, tool use, orchestration, memory, and evaluation
- Evals and observability for production AI. You own how we know the system is working and how we catch it when it is not
- Cost and latency management at production volume
- Partnership with AI product on scoping and sequencing features
- Partnership with data engineering on the pipelines, schemas, and data quality your systems depend on
- Technical mentorship of other engineers working on AI-adjacent systems
- Vendor and model evaluation, including POCs, benchmarks, and cost-performance tradeoffs
Must Haves- 6+ years of software engineering experience, with at least 2 years focused on shipping production LLM-based or ML systems.
- Demonstrated experience building and deploying agentic systems, including tool use, orchestration, and multi-step workflows.
- Strong Python proficiency, including production-grade code quality, testing, and deployment practices.
- Hands-on experience with LLM APIs (Anthropic, OpenAI, AWS Bedrock, or similar), including prompt design, structured outputs, and function calling
- Production experience with evals and observability for LLM systems. You know how to measure accuracy, detect regressions, and monitor drift.
- Experience with retrieval systems (RAG), vector databases, and embedding models.
- Fluency with cloud infrastructure (AWS preferred), including serverless, containers, and API design.
- Clear written and verbal communication skills with the ability to write design docs,.explain tradeoffs, and collaborate with product and engineering peers.
- Ownership mindset - Ships work end-to-end rather than handing off.
Nice to Haves- Fintech, mortgage, or regulated industry experience
- Experience with document AI, OCR pipelines, or structured extraction workflows
- Familiarity with AWS Bedrock, SageMaker, or equivalent cloud AI services
- Experience with modern agent frameworks (LangGraph, CrewAI, AutoGen, custom) and a clear point of view on when each is appropriate
- Experience with modern data warehouses (Snowflake, BigQuery) and transformation tools (dbt)
- Contributions to open source AI or ML projects
- Experience operating production systems on-call
What Success Looks Like
- Within 60 days, you have shipped a meaningful improvement to an existing AI capability or a new one into production
- Within 90 days, evals and observability are in place for the systems you own
- AI systems run reliably at production volume with predictable cost and latency
- Other engineers on the team ship AI work faster and with higher quality because of the patterns and tooling you have established
- You are the person product and engineering leadership trust to give a straight answer on what is shippable, what is not, and what it will take
Salary Band: $90,000 - $150,000 (depending on experience and location)
