Responsibilities
- Design, build, and operate LLM-powered applications, agents, and workflows end-to-end — from prototype to production.
- Architect retrieval, context engineering, and tool-use strategies that make models reliable, accurate, and cost-efficient.
- Integrate LLMs with internal services, third-party APIs, and data stores to automate complex business and engineering workflows.
- Build, evaluate, and continuously improve evaluation harnesses for non-deterministic systems.
- Collaborate closely with product, research, and platform teams to translate ambiguous problems into shipped capabilities.
- Stay ahead of the rapidly evolving LLM ecosystem (models, frameworks, agentic patterns) and bring the best ideas into our stack.
Basic qualifications
- Strong Python skills- you write clean, idiomatic, well-tested code and understand the language deeply.
- Hands-on experience using coding agents(Cursor, Claude Code, GitHub Copilot, or similar) to build complex software systems. You know how to delegate effectively to AI assistants and review their output critically.
- Experience with multiple database paradigms- both SQL (PostgreSQL, MySQL) and NoSQL (MongoDB, Redis, DynamoDB, or similar). You can choose the right tool for the job.
- Experience designing and integrating with third-party APIs- REST and gRPC. Comfortable building robust clients, handling auth, retries, rate limits, and schema evolution.
- Production experience with Docker and Kubernetes- containerizing services, writing manifests, and debugging deployments.
- Strong Linux fundamentals- confident in bash and the terminal; you can navigate, script, and troubleshoot a server without reaching for a GUI.
- Experience building cloud-native tools on AWS, GCP, or Azure (compute, storage, queues, serverless, IAM).
- Solid understanding of what an LLM is and how it works- tokenization, attention, context windows, sampling, and the practical implications of each for system design.
- Strong grasp of modern patterns for integrating LLMs into real workflows, including RAG, MCP (Model Context Protocol), vector databases, agents, tool use, and context engineering- with hands-on experience building with several of them.
- Production experience implementing LLM-powered systems end-to-end, using relevant tools and frameworks (e.g. LangChain, LlamaIndex, LangGraph, Haystack, Pydantic AI, vector stores like Pinecone/Weaviate/pgvector, observability tools like LangSmith or Langfuse).
- Solid foundation in core ML concepts; embeddings, evaluation, overfitting, generalization, and how classical ML relates to and differs from modern LLM-based approaches.
Preferred qualifications
- Experience fine-tuning or distilling open-source models.
- Contributions to open-source AI/ML projects.
- Experience with streaming, real-time systems, or low-latency inference.
- Familiarity with prompt evaluation frameworks and LLM-as-judge methodologies.
- Background in distributed systems or high-scale backend engineering.
- A builder's mindset; you ship, measure, iterate, and don't get stuck in analysis.
- Comfort with ambiguity; LLM systems are non-deterministic and the field moves weekly. You navigate that with curiosity rather than frustration.
- Strong communication; you can explain trade-offs to engineers, product managers, and executives alike.
- Pragmatism; you know when to reach for a 200-line script and when to invest in proper infrastructure.
Tags & focus areas
Used for matching and alerts on DevFound Fulltime Ai Ai Engineer Generative Ai