ML/AI Engineer, Applied AI

We're looking for an ML/AI Engineer, Applied AI to build the AI systems layer behind Drawbridge's process intelligence platform.

This is a full-time role for a senior applied AI engineer who cares about making AI behavior measurable, reliable, safe, and useful in production. We build production AI systems on top of hosted frontier models from providers like OpenAI, Anthropic, Gemini, and Cohere.

You will own the systems that turn messy enterprise data into structured process intelligence: retrieval, context construction, model/provider selection, structured extraction, evals, orchestration, quality loops, and monitoring.

Impact: You'll directly improve the quality, reliability, and usefulness of Drawbridge's AI outputs.

Applied AI ownership: You'll build and own the production systems around LLM APIs.

Evaluation discipline: You'll create the datasets, tests, and quality gates that help us know when the system is improving or regressing.

Product partnership: You'll work closely with backend, product, and forward deployed engineering to solve real customer workflow problems.

Technical leverage: Your work will shape how Drawbridge chooses models, constructs context, evaluates outputs, manages cost/latency, and safely ships AI behavior.

What we're looking for

Must-haves

4-6+ years building production software, ML systems, or applied AI systems

Evaluation mindset: You know how to design tests, compare outputs, build datasets, measure quality, and prevent regressions

Data quality judgment: You understand that AI quality depends on source data, context, labels, edge cases, and feedback loops

Production reliability: You care about latency, cost, observability, retries, failure modes, and making systems debuggable

Systems thinking: You can reason about pipelines end-to-end, from raw customer data to generated outputs and user-facing product behavior

Comfort with ambiguity: You can turn a fuzzy product goal into experiments, implementation, measurement, and shipped improvements

Practical LLM experience: You've built with hosted LLM APIs, including prompting, structured outputs, embeddings, retrieval, tool use, or workflow orchestration

Strong Python skills: You can build reliable services, data pipelines, eval workflows, and AI tooling in production-quality Python

Strongly preferred

RAG and retrieval: Embeddings, vector search, hybrid search, chunking, ranking, reranking, context assembly, or knowledge-base systems

LLM evals: Golden datasets, LLM-as-judge, human review workflows, regression testing, launch gates, offline and online evaluation

Agent/tool orchestration: Multi-step AI workflows, tool calling, structured outputs, planner/executor patterns, or agent reliability work

Prompt and context optimization: Improving output quality through prompt design, schema design, context selection, and model/provider comparison

Safety and guardrails: Reducing hallucinations, enforcing constraints, handling sensitive data, or designing verification and fallback paths

Cost and latency tradeoffs: Practical experience balancing model quality, speed, reliability, and spend

Nice to have

NLP or document understanding: Extraction, classification, summarization, entity resolution, semantic search, or long-context workflows

Process mining, task mining, or enterprise workflow experience

Multimodal inputs: Experience with documents, screenshots, video, transcripts, or mixed enterprise data sources

Go or backend service experience

Customer-facing AI product experience

AI-assisted development: You already use tools like Claude Code, Cursor, Codex, Copilot, or similar to accelerate your work

Our tech stack

Go, Python, PostgreSQL, Redis/Valkey, React 19 + TypeScript, GCP, Terraform, Docker, Datadog, GitHub

AI systems: hosted LLM APIs (OpenAI, Anthropic, Gemini, Cohere), embeddings and retrieval, structured extraction, agent/tool-calling workflows, eval pipelines, and AI observability