EPAM is a leading global provider of digital platform engineering and development services
Responsibilities
Set architecture for systems that span squads; write the decision down and defend it
Build and tune GPU-based serving — vLLM, TGI or TensorRT-LLM — for latency, throughput and cost per token
Own the reference RAG and agent architecture other teams build against
Define service boundaries, API contracts, event-driven infrastructure and SLO work
Develop offline and online evaluation and regression detection
Requirements
5+ years of hands-on engineering experience with shipped production systems spanning applied AI, AI infrastructure and backend platforms
At least 1 year of relevant leadership experience
Expertise in Python and its production ecosystem, including FastAPI, Pydantic and asyncio
Proficiency in production RAG end-to-end: ingestion, vector search, reranking and grounding evaluation
Background in agent orchestration frameworks such as LangGraph, LangChain, OpenAI Agent SDK or Microsoft Agent Framework
Competency in distributed systems, including idempotency, back-pressure and retry semantics
Knowledge of PostgreSQL at scale, GPU inference and serving using vLLM, TGI or TensorRT-LLM
Familiarity with Kubernetes on Azure
Skills in at least one infrastructure language: Go, Rust, Java or Kotlin
Capability to use agentic coding tooling as a daily workflow
Excellent command of written and spoken English (B2+)
Nice to have: pgvector or Qdrant at production scale, Arabic-language/voice/conversational AI, Azure sovereign cloud, Entra ID, Azure OpenAI, Langfuse, Grafana, Prometheus, Terraform
Conditions
Remote work (Belarus, Georgia, Armenia, Kazakhstan, Kyrgyzstan, Uzbekistan)