Senior AI/ML Engineer

Agentic AI Builder Bedrock AgentCore & Claims Automation

Experience: 6+ yrs AI/ML design & development • 2+ yrs hands-on Agentic AI build

Role fit: Hands-on builder implementing agents, RAG, and event-driven pipelines for the automated claims lifecycle

Domain: Healthcare claims semantics; PHI-safe engineering under HIPAA / SOC 2

Works within: Architect-led pod alongside data scientists, cloud architects, and business analysts

PROFILE

Production-focused AI/ML engineer who builds and operates agentic systems, not just prototypes. Implements agents on Amazon Bedrock and AgentCore Runtime, stands up RAG over Bedrock Knowledge Bases, and wires them into event-driven AWS pipelines with the retry, DLQ, and audit patterns that regulated claims processing demands. Fluent in Python, comfortable across the AWS serverless and container stack, and disciplined about least-privilege security and observability.

WHAT THIS PERSON BUILDS ON THE ENGAGEMENT

Implements agents using Strands SDK (or equivalent) on Bedrock AgentCore Runtime deployment, invocation, Memory, Identity, and Observability

Designs and builds RAG architectures on Bedrock Knowledge Bases and vector stores, tuned for claims-domain semantics

Builds Model Context Protocol (MCP) servers and downstream API aggregation

Implements multi-agent orchestration patterns: sequential, parallel, and supervisor

Instruments agents for tracing, monitoring,



and debugging via AgentCore Observability

Builds event-driven pipelines: async processing, Kafka producers/consumers, DLQs, and retry/backoff

Manages claim state and audit trails in DynamoDB; handles PHI with KMS encryption and zero-trust IAM

CORE TECHNICAL STRENGTHS

AI & agents Amazon Bedrock (model invocation, prompt engineering, RAG) • AgentCore Runtime (Memory, Identity, Observability, Gateway & Registry) • MCP server design • Healthcare/claims prompt design

Backend Python • FastAPI / REST integration • Kafka producers & consumers • Lambda functions • Strands agents

Event-driven Async pipelines • Dead-letter queues • Retry / backoff • SQS • EventBridge • Step Functions

Cloud & infra AWS (EKS, Lambda, S3, SQS, EventBridge, DynamoDB, Aurora RDS, CloudWatch) • Terraform • GitHub Actions CI/CD (dev test prod)

Data & compliance DynamoDB claim state + audit trail • S3 • HIPAA / SOC 2 PHI patterns • KMS • Least-privilege IAM tied to AgentCore Identity

Preferred tooling Kiro spec-driven dev, agent code-gen, IaC generation, CI/CD scaffolding

ENGINEERING SIGNALS

Ships production-grade, observable agents with evaluation and guardrails baked in — not notebook demos

Comfortable owning a slice of the claims pipeline end to end and integrating it into the broader architecture

Security- and compliance-first by habit in a PHI setting

📌 Senior AI/ML Engineer (Chennai)
🏢 Pmr Softtech
📍 Chennai