Agentic AI Platform Architect

Prisma Sync Tech; United States; Hybrid; USD 140,000 - 170,000; Full time

Job summary

A technology firm is seeking an Agentic AI Platform Engineer to build scalable, production-grade infrastructure components. The role involves creating standardized AI constructs, focusing on agent orchestration and MLOps practices. Key responsibilities include designing multi-agent architectures, implementing CI/CD for AI agents, and establishing evaluation pipelines. Candidates should be familiar with enterprise-grade MLOps discipline and have experience in AI engineering.

Responsibilities

• Build reusable Agentic components such as MCP servers and data connectors.
• Design multi-agent architectures and tool invocation frameworks.
• Implement CI/CD for AI agents and manage model lifecycle.

Job description

Role Summary

The Agentic AI Platform Engineer builds scalable, reusable, production-grade agentic infrastructure components. This role focuses on creating standardized Agentic AI constructs — including MCP servers, connectors, agent orchestration templates, memory frameworks, evaluation pipelines, and deployment patterns — with enterprise-grade MLOps discipline.

Key Responsibilities

Agentic Platform Engineering (Build reusable Agentic components):

MCP (Model Context Protocol) servers, MCP registries, MCP gateways

Data Connectors (SQL, Databricks, NetDocs, APIs)

Tool orchestration frameworks

Memory & context management services (Short term, long term memory)

Create standardized agent templates for common agent patterns

LLM & Agent Orchestration (Design & deploy agents:

Multi-agent architectures and design patterns

Design Tool invocation frameworks, memory management framework

Define and implement guardrails and policy enforcement at runtime.

MLOps & AI Engineering

Implement CI/CD for AI agents.

Manage model lifecycle using:

MLflow, Model registry, Version control for prompts and agents

Establish: Evaluation pipelines, Observability, Latency monitoring, Hallucination detection, Security testing