Lead AI Engineer

8+ years software engineering experience

3+ years hands-on ML / AI engineering experience

Strong Python (mandatory)

Proven production experience with LLM systems

Deep understanding of:

- RAG architecture

- Embeddings

- Chunking strategies

- Retrieval optimization

Experience with AWS Bedrock

Experience designing tool-calling LLM systems (MCP or equivalent architecture)

Experience integrating external APIs as agent tools

Experience building REST services (FastAPI or similar)

Experience with vector databases (OpenSearch, Pinecone, etc.)

Experience implementing evaluation frameworks for LLM quality

Experience mitigating hallucinations and prompt instability

Strong understanding of:

- Token economics

- Context window constraints

- Cost-performance tradeoffs

Experience integrating with GitHub and JIRA APIs

Production experience beyond notebooks or PoCs

Job Responsibilities

Architect and implement RAG pipelines on AWS Bedrock

Design knowledge ingestion pipelines (JIRA, GitHub, Confluence, S3)

Define chunking, embedding, and retrieval strategies

Design vector storage and retrieval architecture

Architect tool-calling agent systems (MCP or equivalent)

Define when to use:

- Direct API invocation

- Lambda-based tools

- MCP-based tool integration

- Design and implement structured output agents

Implement hallucination mitigation strategies

Build evaluation pipelines for model quality and regression testing

Optimize token usage and latency

Define model routing strategies (cost vs quality)

Implement guardrails and structured validation

Work closely with DevOps to productionize AI services

Mentor AI engineers and define AI engineering standards

Define long-term GenAI roadmap and architecture patterns

Department/Project Description

We are building an AI-driven SDLC automation platform leveraging AWS Bedrock, knowledge bases, domain agents, and tool-based orchestration.
We are looking for a Lead AI Engineer to design and implement the AI architecture, including RAG pipelines, agent systems, tool-calling mechanisms (MCP-style architectures), model evaluation, and guardrails.
This role requires real production GenAI experience, not experimentation.