AWS DE

Job Requirements Primary Stack: Python + FastAPI; Postgres, DynamoDB, Redis; SQS/SNS on AWS; integrations with OpenAI, Anthropic, AWS Bedrock, Azure OpenAI, and custom models; vector stores. Java and data-engineering experience are pluses.

Build FastAPI services that integrate pre-built models (OpenAI, Anthropic, Bedrock, Azure OpenAI, custom) into content and data processing pipelines.

Design prompt engineering, structured-output extraction, retry / fallback strategies, and handling of the many ways model calls can fail (timeouts, rate limits, malformed outputs, hallucinated fields).

Build evaluation harnesses for model outputs — offline eval sets, production sampling, drift detection, and the discipline to know when a prompt or model change has actually improved things.

Handle async processing at scale — SQS/SNS, job queues, DLQs, idempotent retries — for pipelines that call models across large content volumes.

Integrate with vector stores — embedding ingestion, index management, semantic retrieval, hybrid keyword + vector patterns.

Build observability specific to AI systems — latency percentiles by model, error rates, cost tracking, output quality signals, trace correlation across model calls and downstream effects.

Work with data engineers on the pipeline side and with D&A on the model-selection side; keep the seams clean.

Ship with tests, CI/CD, on-call for the services you own.

Use AI coding agents actively as part of your workflow — always with review.

Must-have skills

Python at production depth — you have shipped and operated services that other people depend on. FastAPI or equivalent framework.

Real production experience integrating pre-built models into a product or pipeline — LLMs (OpenAI, Anthropic, Bedrock, or Azure OpenAI), embedding models, or classical models served via API. Not a course project, not a prototype, but something that ran and had real users or real content flowing through it.

REST API design: contracts, validation, structured errors, timeouts, pagination, versioning.

Production AWS experience — IAM, service selection (Lambda, ECS/EKS, S3, SQS), secrets management, cost awareness. Equivalent depth in another major cloud (GCP, Azure) is acceptable; willingness to work in AWS is required.

Work Experience 7-10Years