Lead AI Engineer with .NET

EPAM Systems; United States; Remote; USD 140,000 - 210,000; Full time

Job summary

EPAM Systems seeks a senior engineer to architect and ship AI features such as natural-language query, summarization, and intelligent validation across products. You will establish reusable patterns for retrieval and guardrails, and drive RAG workflows with embeddings and citations.

You will lead a small team on LLM integration, define evaluation methods, and ensure cost, latency and reliability through caching and telemetry. Strong AI governance is required.

Qualifications

• 8+ years of software engineering experience in .NET (C#) and/or Python with lead-level ownership of AI feature architecture across one or more products.
• Proven background in shipping LLM-based features to production, including tool and function calling, structured outputs and streaming.
• Proficiency with Azure OpenAI / Azure AI Foundry, OpenAI or Anthropic APIs.
• Solid SQL fundamentals and API integration skills.

Responsibilities

• Own the architecture of AI features and ship them, including NLP queries, summarization and intelligent validation.
• Establish reusable AI feature patterns for retrieval, evaluation and guardrails to avoid reinvention.
• Drive end-to-end RAG workflows with embeddings, hybrid search and citations.
• Lead and mentor a small engineering team on LLM integration and evaluation.
• Define evaluation discipline with golden datasets and regression checks.

Skills

AI feature architecture

LLM integration

Azure OpenAI

Python

C#

Agentic automation

Lead ownership

Tools

Azure OpenAI API

OpenAI API

Anthropic API

GitHub Copilot

Job description

Responsibilities

Own the architecture of AI features and ship them, including natural-language query, summarization, intelligent validation and exception handling, and workflow assistance Establish and evangelize reusable AI feature patterns for retrieval, evaluation and guardrails so teams do not reinvent them per product Implement RAG end to end: chunking, embeddings, hybrid search, reranking, and grounded responses with citations Drive prompt and context engineering, including multi-step and agentic flows where they add product value Define evaluation discipline, including golden datasets, offline eval suites, LLM-as-judge approaches and per-release regression checks Define and document complex requirements with stakeholders across product features, evaluation criteria and responsible-AI constraints Lead and mentor a small engineering team on LLM integration and evaluation, taking accountability for their work Manage cost, latency and reliability through caching, fallbacks, token budgets and graceful degradation Integrate with platform services such as model gateway, prompt registry, vector stores and embedding pipelines Apply responsible-AI practice, including tenant data isolation, auditability, OWASP LLM Top 10 mitigations and human-in-the-loop patterns Instrument telemetry via Application Insights and Serilog.

Requirements

8+ years of software engineering experience in .NET (C#) and/or Python, with lead-level ownership of AI feature architecture across one or more products Proven background in shipping LLM-based features to production, including tool and function calling, structured outputs and streaming Proficiency with Azure OpenAI / Azure AI Foundry, OpenAI or Anthropic APIs Working knowledge of the component parts of a modern AI/data platform and how to build against them: model gateway, prompt registry, vector stores Familiarity with embedding pipelines, evaluation frameworks, LLM observability and guardrails Judgment about where AI adds product value and where deterministic logic is the better tool Solid SQL fundamentals and API integration skills Strong documentation and standards habits, including architecture docs in Azure DevOps Wiki, code review and testing discipline Hands‑on experience using AI coding agents in the SDLC such as GitHub Copilot, Claude or equivalent Capability to work in agentic automation pipelines, including AI-driven PR review and QA acceptance flows triggered by ADO work item tags English proficiency at an Upper‑Intermediate level (B2) or higher