Forward Deployed AI Engineer - Agentic Systems Lead

EPAM Systems; United States; On-site; USD 180,000 - 260,000; Full time

Job summary

EPAM Systems is seeking an AI-Native Delivery Manager to own end-to-end delivery of production AI-native solutions embedded in client environments. You will design agentic frameworks, operate tooling and evaluations, and drive delivery with stakeholders across business and tech teams.

You will mentor engineers, define realistic timelines, and ensure security and data constraints are addressed while delivering scalable outcomes for clients.

Qualifications

• 8+ years of software engineering with leadership of agentic/LLM-based systems in production.
• Expertise in agent architecture, orchestration, goal decomposition, multi-agent coordination.
• Grounding models in live data via retrieval and context construction.
• Design guardrails, evaluations, and observability for agentic failures.
• Understand production trade-offs: latency, cost per call, rate limits, versioning, SLAs.
• Ownership of technical workstreams with client-environment integration and security constraints.

Responsibilities

• Own technical outcomes end-to-end as the AI-Native Delivery Manager's direct counterpart, driving workstream completion beyond task execution
• Design, build, and operate agentic frameworks, tools, evaluations, and infrastructure that power production AI-native solutions
• Work embedded with client business and technical stakeholders, adapting agentic systems to real integration, data, and security constraints
• Partner with delivery managers on scope and SLA definition, providing technical grounding for realistic commitments and timelines
• Contribute engagement learnings to the delivery playbook, documenting staffing patterns, tooling choices, and integration insights for future teams
• Mentor less experienced engineers on the FDE pillar spectrum, helping them grow from build-focused roles to full delivery ownership

Skills

Software engineering leadership

Agentic/LLM systems

Agent architecture & orchestration

Grounding models in live data

Guardrails, evaluations, observability

Production trade-offs (latency, cost)

Delivery ownership & accountability

Job description

Step into a high-impact role where you'll own end-to-end delivery of AI-native solutions embedded directly within client environments. As a Forward Deployed Expert , you'll design and operate agentic frameworks that solve real business problems, not just build prototypes. Join EPAM to engineer solutions that matter. From AI to cloud transformation, you'll collaborate with top-tier innovators, gain autonomy to explore your ideas, and grow your skills in a culture built for tech excellence.

Req.#1100243513

Responsibilities

• Own technical outcomes end-to-end as the AI-Native Delivery Manager's direct counterpart, driving workstream completion beyond task execution
• Design, build, and operate agentic frameworks, tools, evaluations, and infrastructure that power production AI-native solutions
• Work embedded with client business and technical stakeholders, adapting agentic systems to real integration, data, and security constraints
• Partner with delivery managers on scope and SLA definition, providing technical grounding for realistic commitments and timelines
• Contribute engagement learnings to the delivery playbook, documenting staffing patterns, tooling choices, and integration insights for future teams
• Mentor less experienced engineers on the FDE pillar spectrum, helping them grow from build-focused roles to full delivery ownership

Requirements

• 8+ years of software engineering experience with recent hands-on leadership of agentic or LLM-based systems in production environments
• Deep expertise in agent architecture and orchestration, including goal decomposition, multi-agent coordination, and autonomy-versus-handoff decision frameworks
• Proven experience grounding models in live data systems through retrieval, context construction, and maintaining accuracy against changing knowledge sources
• Ability to design guardrails, evaluations, and observability for agentic failure modes including hallucination, tool-call errors, and runaway loops
• Understanding of production trade-offs such as latency, cost per call, rate limits, versioning, and rollback under fixed-fee or managed-output SLAs
• Demonstrated ownership of technical workstream outcomes with accountability for results, including integration into client environments with security and legacy constraints