Lead Autonomous Agent Architect

EPAM Systems; Chile; On-site; CLP 12,000,000 - 18,000,000; Full time

Benefits offered by this job

Healthcare benefits

Paid time off

Upskilling programs

Global career opportunities

Volunteer opportunities

Learning library access

Job summary

EPAM Systems seeks a Chief Agentic AI Engineer to architect, deliver, and run autonomous agent systems at scale with production-grade reliability. You will shape end-to-end agentic workflows across runtime, memory, identity, tools, and observability while raising the bar for evaluation, tracing, and guardrails.

Responsibilities include defining architecture for agents, championing Bedrock AgentCore adoption, designing memory systems, building MCP servers and tool integrations, and optimizing

Qualifications

• 7+ years of software development experience.
• 2+ years shipping LLM-based agents to production.
• Proficiency in Python and agent frameworks.

Responsibilities

• Define architecture for autonomous agents across runtime, memory, identity, tools, and observability.
• Champion adoption of AWS Bedrock AgentCore and Strands Agents SDK in production.
• Design long-term memory and context-management for multi-step engagements.
• Create and maintain MCP servers and tool integrations for agents.
• Improve reliability, cost, and latency of agent reasoning with standards for evals, tracing, and guardrails.

Skills

Python

LLM Agents

Agent frameworks

Observability

Security boundaries

Tools

AWS Bedrock AgentCore

LangGraph

Claude Agent SDK

OpenAI Agents SDK

CrewAI

Job description

EPAM Systems seeks a Chief Agentic AI Engineer to architect, deliver, and run autonomous agent systems at scale with production-grade reliability. You will shape end-to-end agentic workflows across runtime, memory, identity, tools, and observability while raising the bar for evaluation, tracing, and guardrails.

Responsibilities include defining architecture for agents, championing Bedrock AgentCore adoption, designing memory systems, building MCP servers and tool integrations, and optimizing