Apply on Employer Site

Codoxo · 1 month ago

AI and Agentic Product Engineer – Healthcare Payment Integrity

United States

Full-time

Remote

Mid Level

3+ years exp

Codoxo is the premier provider of artificial intelligence-driven solutions that help healthcare companies and government agencies proactively detect and reduce risks from fraud, waste, and abuse. As the AI and Agentic Product Engineer, you will be responsible for designing, delivering, and operating agentic AI products that accelerate claim audits and SIU investigations.

AnalyticsArtificial Intelligence (AI)Fraud DetectionGenerative AIHealth CareMachine LearningSoftware

Growth Opportunities

Responsibilities

Build serverless agentic product architectures and features on AWS using Bedrock AgentCore Memory, Gateway, Identity, Interpreter, Observability and Runtime

Design short‑ and long‑term conversational memory for claim, fraud analyst and provider contexts

Implement and operate A2A/MCP servers on AWS; integrate with Bedrock Agents/Converse APIs

Orchestrate multi‑agent plans with Strands Agents; deploy to EKS or Lambda

Add RAG/knowledge bases for claim history, guidelines, and policy context

Explore and implement serverless friendly AI/ML techniques (LoRA/QLoRA) and PEFT (Prompt-Efficient Fine-Tuning) to enhance model performance and deployability

Convert APIs/Lambda into MCP tools with Gateway; secure, version, and catalog them

Instrument agents with CloudWatch metrics, spans, and traces; enable audits

Ship measurable impact

Qualification

PythonGenAI/LLM/AgentsAWSBedrock AgentCoreMCP ServersStrands AgentsData systemsNode.jsGoKubernetes/EKS

Required

3+ years top Python programming skills required

2+ years GenAI/LLM/Agents

1+ years hands‑on with Bedrock AgentCore Memory, Gateway, and Runtime

Proficient with Amazon MCP Servers, Bedrock Agents, Runtime & Strands SDKs

Data systems: S3, Glue/Athena

Experience integrating MCP tools and operating MCP servers on AWS

Practical with Strands Agents (multi‑agent orchestration, tools, handoffs)

LLM and agent tuning/evals: PEFT (LoRA/QLoRA), embeddings, prompt‑ops, human-in-the-loop safety guardrails

Event‑driven AWS: Lambda, Step Functions, A2A, EventBridge, API Gateway

Vector search (FAISS/OpenSearch/pgvector), RAG, and knowledge bases

Healthcare data: FHIR, ICD‑10, CPT, HCPCS; payer claims experience

Physical Requirements: Work is performed in an office environment (either in our office or work-from home) and requires the ability to work on a computer, operate standard office equipment, and work at a desk