Job Description & Details
This is a hands-on technical contract to build and scale agentic AI workflows and contact center solutions directly on AWS. If you are tired of theoretical AI slide decks and want to actually ship production-grade RAG pipelines, LLM agents, and Amazon Connect integrations, this project gives you the greenfield architecture to do it.
What You'll Actually Be Doing
Your day-to-day will revolve around architecting Python services with FastAPI that orchestrate large language models, tools, and vector databases. You'll spend significant time wiring up AWS Bedrock, optimizing OpenSearch retrieval pipelines, and configuring Amazon Connect contact flows so that AI can genuinely assist or automate support interactions. Expect to spend a lot of time wrestling with prompt templates, structured JSON outputs via Pydantic, and keeping serverless latency low on Lambda and Fargate.
The Core Tech Stack
You need deep fluency in Python for orchestration alongside frameworks like LangChain or Semantic Kernel. On the AWS side, mastery of Bedrock, OpenSearch, Lambda, Fargate, and CloudFormation is non-negotiable since you will be deploying and managing these components yourself. Experience with Amazon Connect is a major differentiator here, as the project heavily relies on integrating generative AI directly into contact center workflows.
Interview Expectations
Expect the engineering team to grill you on how you handle latency and hallucinations in production RAG systems. They will likely ask you to explain how you would design a multi-agent workflow that fails gracefully when an LLM times out or returns malformed JSON, and they'll want concrete examples of how you tune OpenSearch vector schemas for hybrid search. They aren't looking for high-level definitions; they want to know how you debug asynchronous Python API bottlenecks and secure endpoints using IAM least privilege.
Application Advice
When updating your resume for this role, make sure your Python backend experience and AWS infrastructure work are front and center. Explicitly highlight terms like OpenSearch, Bedrock, FastAPI, and CloudFormation so you easily clear the ATS keyword filters. Rather than just listing tools, write bullet points that emphasize metrics like cost optimization, latency reduction, and successful deployment of LLM guardrails in production.