full stack developer for agentic systems

Описание:

EPAM develops enterprise software products, open source solutions, and accelerators.

Задачи:

Industrialize AI prototypes into production services on Azure AI Foundry and Azure OpenAI; Build agentic systems using orchestration frameworks, RAG pipelines, vector databases, knowledge graphs, and MCP-based tool integration; Establish evaluation harnesses with golden sets, regression evaluations, and guardrail tests integrated into CI; Engineer guardrails for input and output filtering, grounding and citation, PII protection, rate limits, and human escalation paths; Deliver full-stack solutions across Python, Java Spring Boot services, and TypeScript/Angular front ends; Run production engineering end-to-end, including CI/CD, observability, cost management, and latency management; Design systems to remain resilient as underlying models evolve; Build security and compliance into prompts, stores, and logs, externalize secrets, and ensure AI decisions are auditable; Iterate based on real usage through hypercare, tuning, and fixes, then package successful patterns for future use.

Требования:

5+ Years of experience shipping LLM and agentic systems in production with real users, defined evaluation approaches, and demonstrated scale; Proficiency in Python, Java Spring Boot, and TypeScript/Angular for full-stack delivery; Expertise in Azure PaaS, Azure AI Foundry, and Azure OpenAI services; Knowledge of agentic frameworks, vector databases, knowledge graphs, and MCP-based tool integration; Skills in prompt and context engineering; Experience with CI/CD pipelines and observability practices while operating production systems end-to-end; Familiarity with AI-assisted engineering workflows, including daily coding with AI agents and live workflow demonstration; Understanding of evaluation-first development with measured baselines and regression evaluations; Competency in cost and latency optimization, including token economics; Understanding of regulated-environment practices, including data classification, auditability, and least privilege principles; English proficiency at B2 level or higher.

Условия:

Office work in Warsaw is required 3 days per week.