Senior AI Analytics Engineer
Summary
Hands-on engineer who spends the day writing and shipping production Python/TypeScript code for an AWS-based conversational AI platform: building Bedrock agents, RAG and extraction pipelines (Lambda, Step Functions, DynamoDB), natural-language-to-SQL, and graph queries, plus iterating on prompts, running evals, and debugging production issues.
About the role
You'll spend essentially all day writing code and shipping working systems — implementing the conversational AI, extraction pipelines, and pattern-detection capabilities that the Lead and Senior Director design. You'll take well-scoped pieces of the roadmap and turn them into deployed, tested, working features. If what excites you is being deep in an IDE, a notebook, or a terminal — debugging a flaky Step Functions execution, tightening a prompt until it stops hallucinating, or shaving latency off a retrieval call — this role is built for you.
Key responsibilities
Write, test, and ship production code (Python and/or TypeScript/Node.js) for the platform's extraction pipelines, Bedrock agent tools, and retrieval/analytics services.
Implement and iterate on Bedrock Agents, RAG pipelines (Knowledge Bases / OpenSearch), natural-language-to-SQL over Athena, and knowledge-graph queries (Neptune / GraphRAG) from designs handed off by the Lead.
Build and maintain the AWS extraction pipeline: Lambda functions, Step Functions state machines, DynamoDB state tracking, and Bedrock Converse API calls that turn raw unstructured data into structured facts.
Write and refine prompts and structured-output schemas, iterating quickly against real examples until accuracy holds up.
Build and run evaluation scripts and golden test sets; when something fails, dig down to the specific prompt, retrieval, or data issue causing it.
Instrument, monitor, and troubleshoot production pipelines and agents — read the logs, trace the failure, fix the root cause, ship the fix.
Write unit and integration tests, participate in code review, and keep the codebase clean enough for someone else to pick up.
Look for opportunities to cut cost and latency (model tiering, caching, batching) without needing a reminder.
About you
3–6+ years of hands-on software or data engineering experience, writing and shipping production code.
Daily working proficiency in Python and/or TypeScript/Node.js on real production systems.
Practical experience with core AWS services such as Lambda, Step Functions, S3, and DynamoDB.
Some hands-on exposure to building with LLMs — prompting, RAG, or agent/tool-calling patterns — even from personal projects or a previous role.
Comfort with SQL and basic data analysis (Athena, Redshift, or similar).
A genuine debugging mindset: comfortable reading logs, stack traces, and unfamiliar code to find the actual root cause rather than the first plausible one.
Bachelor's degree in Computer Science, Engineering, or a related field, or equivalent practical experience.
Preferred: Direct experience with Amazon Bedrock (Converse API, Knowledge Bases, Agents/AgentCore).
Preferred: Experience with a vector database (OpenSearch Serverless or similar) or a graph database (Amazon Neptune).
Preferred: Experience with Amazon Comprehend or other NLP/document-processing tooling.