Chief Forward Deployed Engineer

Our requirements

• 7+ years of engineering experience, with a strong recent track record building production AI / LLM applications (not prototypes or research only)
• Strong agent-design judgment — task-harness fit, matching the harness to the context, failures, and policies of the actual task rather than calling a model in a loop
• The ability to operate close to the client: lead discovery and feasibility conversations, work directly with SMEs and end-users, and explain technical trade-offs to both technical and non-technical audiences
• Hands-on experience with agentic frameworks (LangChain, LangGraph, Semantic Kernel, or similar) and major LLM providers (OpenAI, Anthropic, Google Gemini)
• Expert-level Python and solid software engineering fundamentals
• Strong RAG and retrieval skills: vector databases, embeddings, hybrid search, re-ranking, chunking, and context management
• Proven experience evaluating generative AI quality — LLM-based evaluation, heuristics, custom eval frameworks — and using observability/tracing tools (LangSmith, Arize Phoenix, Langfuse, or similar)
• Production deployment experience on at least one major cloud (AWS, Azure, or GCP) with containerization, CI/CD
• Sound judgment under ambiguity — scoping, sequencing, and making the call on speed vs. quality vs. scope
• English at C1 level

Optional

• Experience designing experiments, A/B testing, and iterating on AI products against real user behavior and business metrics
• Background in NLP, Data Science, or applied ML, with experience moving models into production
• Familiarity with MCP, A2A, Agent Skills, and emerging agent standards
• Experience with enterprise AI platforms (AWS Bedrock AgentCore, Databricks Genie, Microsoft Foundry, Gemini Enterprise)
• Exposure to AI governance, security, and compliance (guardrails, prompt-injection prevention)
• Prior client-facing or pre-sales exposure in a consulting or services context

Your responsibilities

• Design, build, and ship AI-native systems E2E — agents, workflows, RAG, and the harness: custom tool calling, sandboxing, context engineering and sub-agents, caching, compaction
• Build the evaluation pipelines and use them to prove the system is genuinely useful
• Design for failure in the agent loop: retries, model fallbacks, cost limits, and human-in-the-loop on consequential actions
• Capture domain expertise and repeatable workflows — so what works on one engagement carries to the next
• Engage early, to help shape the use case and check technical feasibility
• Write production-grade Python: integrations, APIs, data access, deployment
• Work directly with SMEs and end-users — interviews, UAT, observing the real workflow — and validate that the system fits how people actually work

About the project

This is how we organize our work

This is how we work on a project

• Continuous Deployment
• Continuous Integration

What we offer

Development opportunities we offer

• conferences in Poland
• development budget
• external training
• industry-specific e-learning platforms
• intracompany training
• mentoring
• soft skills training
• space for experimenting
• substantive support from technological leaders
• support of IT events
• technical knowledge exchange within the company

Benefits

• private medical care
• life insurance
• flexible working time
• fruits
• integration events
• retirement pension plan
• no dress code
• video games at work
• coffee / tea
• parking space for employees
• leisure zone
• employee referral program
• charity initiatives

Recruitment stages

• 1.
  Recruiter Interview (30 min)
• 2.
  Technical Interview (90 min)
• 3.
  Manager Interview (60 min)
• 4.
  Meeting with the Project Team or Client (60 min, optional)
• 5.
  Welcome Aboard!

EPAM Systems (Poland) sp. z o.o.