[Hiring] Python AI Engineer @EPAM Systems

Role Description

We are hiring a Python AI Engineer to operate and evolve the infrastructure behind an enterprise knowledge base platform as it transitions from a Confluence-focused RAG chatbot into a broader, agentic knowledge system.

Today, our platform supports:

• Confluence/GitHub ingestion
• Chunking
• pgvector
• RAG retrieval
• FastAPI serving

In our next growth phase, we are expanding toward:

• Hybrid retrieval (vector + sparse + graph)
• Multi-source ingestion
• Robust evaluation pipelines
• Agent infrastructure
• Shared chat platform primitives

If you are passionate about building scalable, reliable, and production-grade AI infrastructure rather than just chasing demo behavior, we want to hear from you.

Qualifications

• Strong, hands-on Python experience in platform, automation, or infrastructure-heavy environments
• Experience building CLI tools using Python, Golang, or Rust
• Hands-on experience with LangGraph, LangChain, pgvector, and modern retrieval pipelines
• Experience designing evaluation frameworks for LLM-backed systems, including regression detection and quality measurement
• Strong experience with Docker, Helm, GitHub Actions, and Kubernetes-oriented workflows
• Familiarity with the operational characteristics of embedding pipelines, vector search, and LLM-backed systems
• Strong skills across metrics, tracing, dashboards, alerting, and log analysis
• Experience with ingestion, ETL, or content-processing pipelines at scale
• An innate ability to think in terms of reliability, cost, latency, throughput, and disaster recovery

Requirements

• Architect and implement robust backend services using Python 3.11, FastAPI, Pydantic, SQLAlchemy async, and asyncpg
• Design trade-offs around retrieval quality, latency, cost, safety, and operational simplicity
• Build production-grade agent runtime capabilities, including memory boundaries, tool sandboxing, permissions, and budget controls
• Improve answer grounding, failure analysis, and citation enforcement for enterprise-grade dependability
• Create comprehensive observability and operational feedback loops using OpenTelemetry, Prometheus/Grafana, Docker/Helm, and GitHub Actions
• Ingestion infrastructure across current and future content sources
• Full-stack observability across application, pipeline, database, and model-serving behavior
• Cost, latency, throughput, and failure-mode management for AI-heavy workloads
• Release workflows that validate AI behavior changes, not just code compilation
• Partner closely with product and engineering teams to support multiple conversational surfaces through a unified knowledge platform

Company Description