Senior Agentic AI Engineer: Deploy Production-Grade Agents

EPAM Systems, Inc.; México; A distancia; MXN 1.200.000 - 1.300.000; Jornada completa

Descripción de la vacante

EPAM Systems, Inc. is seeking a Senior Agentic AI Engineer to design, build, and operate autonomous agent systems at scale. You will architect runtime, memory, identity, tools, and observability, and lead deployments using AWS Bedrock AgentCore and Strands SDK across multiple agent harnesses.

You will shape long-term memory, MCP servers, and tool integrations, driving reliability, cost, and latency while establishing guardrails and evals for safe agent reasoning.

Formación

• 3+ years of experience building production software.
• 1+ years shipping LLM-based agents to production.
• Proficiency in Python and hands-on experience with an agentic framework.
• Familiarity with AWS Bedrock AgentCore primitives (Runtime, Memory, Identity, Gateway, Observability).
• Understanding of MCP, tool use, prompt/context engineering, memory architectures, and agent evaluation.
• Skills in production design: observability, guardrails, retries, cost control, and security boundaries.

Responsabilidades

• Design the architecture for building, deploying, and operating autonomous agents at scale.
• Lead adoption of AWS Bedrock AgentCore and Strands Agents SDK across deployments.
• Design long-term memory and context-management across multi-step engagements.
• Build MCP servers and tool integrations used by agents (recon, scanners, exploit primitives, internal services).
• Drive reliability, cost, latency, evals, tracing, and guardrails for agent reasoning.

Conocimientos

Python

Agent frameworks

AWS Bedrock

Observability

Memory architecture

Security boundaries

Herramientas

LangGraph

Claude SDK

OpenAI Agents SDK

CrewAI

AgentCore

Descripción del empleo

EPAM Systems, Inc. is seeking a Senior Agentic AI Engineer to design, build, and operate autonomous agent systems at scale. You will architect runtime, memory, identity, tools, and observability, and lead deployments using AWS Bedrock AgentCore and Strands SDK across multiple agent harnesses.

You will shape long-term memory, MCP servers, and tool integrations, driving reliability, cost, and latency while establishing guardrails and evals for safe agent reasoning.