LLM System Design & Deployment

• Design and implement LLM-powered features end-to-end — from prompt architecture and model selection through API integration and production deployment — with minimal supervision.
• Own prompt engineering for production features: design, version, and systematically evaluate prompts across model updates and behavior regressions.
• Integrate conversational and agentic AI capabilities into an existing application, owning the API layer, session management, and graceful degradation strategies.
  
  

RAG & Retrieval Systems

• Build and maintain RAG pipelines — including chunking strategy, embedding selection, vector store management, and retrieval evaluation — tuned for the application's domain.
• Work across retrieval approaches (dense vector search, BM25 hybrid, re-ranking) and evaluate trade-offs for accuracy, latency, and cost.
  
  

Agentic Workflows & Orchestration

• Select and apply frameworks (LangChain, LlamaIndex, LangGraph, custom) based on real trade-offs in the context of the product — not hype.
• Build with and extend MCP (Model Context Protocol) servers for tool integration, external service access, and structured agent communication.
  
  

Evaluation & Quality

• Define and run LLM evaluation pipelines — automated metrics, human eval, regression suites — and act on results without waiting for direction.
• Identify prompt regressions, retrieval quality issues, and latency problems early and drive resolution.
  
  

Collaboration & Engineering Culture

• Collaborate with backend and frontend engineers as a peer, translating AI capabilities into clean service contracts and integration specs.
• Identify architectural or data quality issues early and escalate when scope warrants.
• Stay current with the LLM ecosystem and bring concrete, well-reasoned proposals for adopting techniques or tooling that address real product problems.
• Contribute to technical documentation, internal best practices, and code reviews for junior team members.

• BSc or MSc in Computer Science, Machine Learning, AI, or a related field.
• At least 1–2 years of hands-on experience in LLM engineering — through industry, coursework, or substantive personal projects.
• Solid understanding of transformer-based LLM architectures and how model behavior, context windows, and inference parameters affect output.
  
  

AI / ML Expertise

• Practical experience building RAG pipelines: chunking, embedding models, vector stores (Pinecone, Weaviate, pgvector, Chroma), and retrieval evaluation.
• Familiarity with agentic frameworks and orchestration patterns: tool use, memory systems, multi-step reasoning, and agent-to-agent communication.
• Understanding of MCP (Model Context Protocol) for building interoperable tool integrations and structured agent workflows.
• Experience with LLM tooling such as LangChain, LlamaIndex, LangGraph, or equivalent — with an ability to go beyond the framework when needed.
• Awareness of prompt evaluation techniques: LLM-as-judge, embedding similarity, regression testing, and structured output validation.
  
  

Engineering Skills

• Strong data preprocessing skills: regex, normalization, pipeline design, and working with messy real-world data.
• Proficiency in Python, with exposure to REST API design and async patterns.
• Familiarity with containerization (Docker) and cloud deployment on Azure.
• Comfort working in a codebase with legacy components and the judgment to integrate cleanly without over-engineering.