Tech Stack
AzureCloudDjangoFlaskGrafanaJavaScriptNext.jsPrometheusPythonReact
About the role
- Build and maintain backend services using Azure Functions, Semantic Kernel, LangChain/LangGraph/AutoGen/CrewAI.
- Develop APIs to orchestrate LLM workflows, manage state, and support chat and analytics front ends.
- Implement multi-agent pipelines with planning, reasoning, and execution flows.
- Design and manage RAG pipelines (chunking, embedding, indexing, hybrid retrieval) and extend retrieval with GraphRAG and entity-driven reasoning.
- Build ingestion pipelines to validate, clean, and test data before adding it to the AI knowledge layer; integrate data from CosmosDB, Dataverse, APIs, Azure Cognitive Search, SharePoint, and other enterprise sources.
- Create automated evaluation methods for retrieval accuracy, hallucination, and dataset quality.
- Collaborate with front-end engineers to define API contracts, payloads, session flows; provide reusable backend services, SDKs, and documentation aligned with React/Next.js and shadcn frameworks.
- Implement telemetry, logging, and metrics (OpenTelemetry, Prometheus, Grafana); enforce RBAC, compliance, data masking/sanitization/lineage; optimize performance, latency, and token usage with robust error handling and scalable design.
Requirements
- Strong hands-on experience with Azure AI/data stack: Cognitive Search, Azure OpenAI, CosmosDB, Azure Functions, AKS, Dataverse.
- Proficiency in Python and backend frameworks (FastAPI/Flask/Django).
- Experience with RAG pipelines, multi-agent frameworks (LangChain, LangGraph, AutoGen, CrewAI), and Semantic Kernel.
- Familiarity with vector stores (FAISS, Pinecone, Weaviate, Azure Cognitive Search).
- Working knowledge of React/Next.js to support API integration with UIs.
- 4+ years in backend, ML infrastructure, or cloud engineering.
- 2+ years building AI/LLM-based applications, ideally in enterprise environments.
- Proven experience integrating multiple enterprise data sources securely.
- Background in data quality testing, validation, and continuous evaluation for AI systems.
- Exposure to full-stack development with React/Next.js.
- Nice-to-haves: Experience with GraphRAG, knowledge graphs, continuous LLM evaluation (DeepEval, G-Eval), end-to-end full-stack AI solutions.