Python AI Engineer for Generative AI & RAG Systems

Deloitte (UK); Alabama; On-site; USD 120,000 - 180,000; Full time

Job summary

Deloitte invites a hands-on software engineer to join its Associate (Contractor) Programme in a fast-moving Generative AI environment. You will develop production-grade Python services for AI workflows, focusing on modularity, reliability and scalability across LLM integration and RAG systems.

Collaborating with engineering, data, platform and evaluation specialists, you will implement multi-stage retrieval, structured outputs, tests, logging and observability, while translating evaluation

Qualifications

• Production-grade Python development experience.
• Hands-on Generative AI and RAG delivery beyond PoC.
• Experience with APIs, testing and deployment.

Responsibilities

• Design, develop and maintain production-grade Python services for Generative AI and complex RAG solutions.
• Define typed application, API and model contracts.
• Integrate LLMs using structured outputs, tool calling, schema validation, context management and controlled fallback mechanisms.
• Implement multi-stage retrieval using vector search, lexical search, hybrid retrieval, metadata filtering, reranking and evidence selection.
• Build automated tests (unit, integration, contract, regression).
• Implement structured logging, tracing and operational metrics across AI workflows.

Skills

Production-grade Python

Async programming

Type annotations

LLM integration

Testing with Pytest

CI/CD

Docker

Kubernetes

Azure OpenAI

Knowledge graphs

Tools

LangGraph

Semantic Kernel

Pydantic AI

Azure Foundry

Azure AI Search

Job description

Deloitte invites a hands-on software engineer to join its Associate (Contractor) Programme in a fast-moving Generative AI environment. You will develop production-grade Python services for AI workflows, focusing on modularity, reliability and scalability across LLM integration and RAG systems.

Collaborating with engineering, data, platform and evaluation specialists, you will implement multi-stage retrieval, structured outputs, tests, logging and observability, while translating evaluation