Leads the team you would join
Signs off on engineering hires
Owns the opening
From your own Gmail and LinkedIn.
About the role
This is a senior role for an engineer indexed as based in Pune, who designs and ships production agentic AI systems on AWS Bedrock. The work covers multi-agent orchestration, retrieval-augmented generation pipelines, and the memory, tool-calling and safety layers around them. The employer expects three to five or more years overall, with one to two years of production LLM and agentic work, and wants hands-on delivery rather than prototypes.
What you’ll do
- Design multi-agent workflows with planners, executors, retrievers, validators and human-in-the-loop steps.
- Implement reasoning patterns with safety and observability controls.
- Build memory systems using state graphs, session logs and long-term knowledge bases.
- Develop task execution engines with tool calling, API integration, fallbacks and escalation.
- Deploy and scale foundation models through AWS Bedrock, including Knowledge Bases, Agents, Guardrails and evaluations.
- Build serverless backend workflows with Lambda and Step Functions.
- Build ingestion pipelines for embeddings, chunking, metadata and semantic indexing.
- Set up LLMOps practices for prompt versioning, logging, latency, quality and cost.
- Design evaluation datasets to track drift and response quality.
- Review code and document architecture, workflows and operating procedures.
What they’re looking for
- Expert Python skills and asynchronous APIs with FastAPI, Flask or similar
- Hands-on AWS Bedrock experience with models such as Claude, Llama and Mistral
- Experience with EC2, Lambda, IAM, VPCs and Step Functions
- Experience with agent frameworks such as LangGraph, CrewAI, AutoGen or Semantic Kernel
- RAG architecture experience across ingestion, retrieval and reranking
- Experience with vector databases such as Pinecone, OpenSearch, pgvector or Chroma
- LLMOps and AI safety knowledge, including guardrails and prompt-injection handling
- Three to five or more years overall, including one to two years of production LLM or agentic work
Nice to have
- Contributions to open-source agent libraries or MCP ecosystems
- PEFT or LoRA fine-tuning experience
- Model monitoring and drift analysis
- Infrastructure as code, containers and CI/CD
- Threat modelling for AI agents
Skills
Questions about this role
Which cloud platform does the role use?
The role centres on AWS Bedrock, with AWS Lambda, Step Functions and related services.
How many years of experience are needed?
Three to five or more years overall, with at least one to two years in production LLM and agentic AI work.
Which frameworks should I know?
Agent frameworks such as LangGraph, CrewAI, AutoGen or Semantic Kernel, plus vector databases such as Pinecone or pgvector.