Senior AI Engineer resume example

• As workstream lead, built a Python RAG service over permission-filtered support documents. Compared retrieval depth and prompt engineering variants using fixed answer and citation tests; selected the configuration that resolved untraceable citations without exposing restricted documents.
• With responsibility for the review standard, added automated testing to CI/CD for refusal behavior, citation validity, and prompt injection; connected monitoring to the same failure categories so a fluent but unsupported answer could block a release.
• As workstream lead, deployed the Python service as a versioned Docker image on AWS ECS, with scoped IAM access to source documents. Validated health checks and rollback against the previous image; released the service only after citation and permission tests passed.
• Built an evaluation set separating retrieval errors from unsupported answers, checked document permissions before retrieval, and recorded citation accuracy alongside latency and token cost.
• Owned a RAG evaluation using synthetic clinical questions and Python fixtures, resolved unsupported citations and completed human review before a limited demonstration.
• Built automated testing for prompt versions using Docker and CI/CD, reduced repeated citation failures by 34% on the held-out fixture and published observability checks.

• Added approval gates and idempotency keys to an agent tool workflow that updates customer records, with a replayable audit log; aligned review standards and decision owners across the participating teams. Working with partner teams, prevented a retried tool call from creating duplicate customer updates.
• Investigated a discrepancy between a published metric and its source records, traced the transformation that changed the population, and corrected the calculation with a reproducible query.
• Compared the last successful data refresh with a failed run, separated missing source data from transformation errors, and reran only the affected interval. Kept the original query and corrected result together for review.

• Added monitoring for token usage and p95 latency by request type, then routed simple requests to a smaller model with an explicit fallback; aligned review standards and decision owners across the participating teams
• Generated synthetic source records with duplicates, late arrivals and corrected values; wrote assertions for row counts and key uniqueness, and recorded the expected effect of each case on the reported metric.
• Compared the analytical output with a manually calculated reference table, traced differences to a transformation step, and retained a data dictionary and rerun instructions alongside the corrected query.
• Owned the synthetic-data validation using a manually calculated reference, resolved duplicate-key inflation and completed a notebook that reproduces the corrected totals.

Relevant coursework: Algorithms, operating systems, databases, computer networks

Python · RAG · LLM evaluation · Vector search · Observability

• Published an independent methods note using reproducible queries, explaining the data grain, excluded records and sensitivity of the result to a changed denominator.