AI/ML Engineer – Generative AI

Key Responsibilities

• Deploy, configure, and extend the AWS GenAI LLM Chatbot solution under an AppSec-approved configuration; adapt it to Client's content, branding, and access model.
• Build ingestion pipelines for policies, procedures, program documentation, and eligibility rules: parsing, chunking, metadata extraction, embedding generation, and incremental refresh.
• Implement retrieval quality work — hybrid search, metadata filtering, re-ranking, query rewriting — and tune against measured thresholds rather than intuition.
• Engineer system prompts and grounding instructions that enforce citation, constrain scope, and resist prompt injection and misuse.
• Implement and validate Amazon Bedrock Guardrails — content filters, PII redaction, denied topics — and produce test evidence that prohibited content is actually blocked before milestone acceptance.
• Implement document-level access control with Amazon Cognito and role-based authorization; write and execute explicit test cases proving users cannot retrieve unauthorized content.
• Build the evaluation harness: golden question sets, groundedness and citation scoring, regression runs on every material change.
• Contribute reusable accelerators to the Enterprise Hub — RAG bootstrap templates, prompt management patterns.
• Support WS#1 by prototyping feasibility spikes for candidate use cases (Policy/Procedure Summarization, Document Intake & Classification, AI BI Analyst via QuickSight Q).
• Write the operational runbook and pair with developers until they can independently redeploy and refresh content.

Required Qualifications

• 4+ years software engineering, with 1.5+ years building LLM-based applications.
• Strong Python and TypeScript; comfortable in a CDK-based codebase.
• Production RAG experience: embeddings, vector databases, chunking strategy, retrieval evaluation.
• Hands-on Amazon Bedrock — model invocation, Knowledge Bases, Guardrails, Agents.
• Working knowledge of Cognito, IAM, and application-layer authorization.
• Demonstrated ability to evaluate LLM output quality with data rather than anecdote.

Preferred Qualifications

• Experience with the aws-samples GenAI chatbot solution or similar reference implementations.
• OpenSearch Serverless, Aurora pgvector, or Kendra experience.
• Document AI experience: Textract, Comprehend Medical, classification and metadata extraction pipelines.
• Amazon Q Business and QuickSight Q exposure.
• Frontend skills (React) sufficient to adapt a chat UI.