Lead Agentic AI Engineer - US

CodeRoad; United States; On-site; USD 190,000 - 250,000; Full time

Benefits offered by this job

Medical insurance

Dental insurance

Vision insurance

Retirement plan

Paid time off

Job summary

CodeRoad is seeking a Lead Agentic AI Engineer to architect and operationalize enterprise-grade autonomous agent systems integrated with Salesforce. You will steer stateful multi-agent workflows using LangGraph, LangChain, and Anthropic Claude, ensuring secure, high-throughput performance within enterprise constraints.

You will define evaluation pipelines, security practices, and delivery standards while collaborating with executives and engineering teams to translate complex AI mechanics into

Qualifications

• 7+ years of hands-on enterprise software engineering experience using Python, TypeScript/Node.js, Java, or Apex.
• 3+ years of direct production experience building, evaluating, and scaling LLM-powered applications and autonomous multi-agent workflows.
• Tech Stack: Expert proficiency with LangGraph / LangChain, Anthropic/Claude APIs (tool calling, prompt caching, Claude Code), MCP, and observability tools LangSmith, Langfuse, or OpenTelemetry.
• Evaluation Frameworks: Hands-on experience with automated LLM eval suites Promptfoo, Ragas, or LLM-as-a-judge patterns as automated test gates.
• Ownership mindset with deep diagnostic debugging skills for isolating non-deterministic LLM behavior, context degradation, and latency bottlenecks.
• Consultative Communication: Strong ability to translate complex AI mechanics into actionable architecture roadmaps for technical and non-technical stakeholders.
• Advanced English: Fluent written and spoken professional English for direct client consultation.

Responsibilities

• Architect stateful multi-agent coordination graphs using LangGraph or custom DAG loops, supporting dynamic task decomposition, supervisor orchestration, human-in-the-loop escalation, and automated error recovery.
• Build standardized MCP (Model Context Protocol) servers and connectors to securely expose Salesforce tools, objects, custom actions, and operational context to agent runners.
• Optimize query routes and context engineering across model tiers (e.g., Claude Sonnet/Opus vs Haiku) to improve latency, accuracy, and token spend.
• Design automated evaluation harnesses (Promptfoo, Ragas, or DeepEval) integrated into CI/CD pipelines to benchmark grounding, schema validity, and hallucination resistance against reference datasets.
• Implement full-stack observability and defense-in-depth security using tools like LangSmith or Langfuse alongside input sanitization, FLS, and PII scrubbing.
• Lead technical delivery standards, establishing API interface contracts, pattern-driven development, and architectural blueprints for cross-functional teams and exec stakeholders.

Skills

Python

TypeScript/Node.js

Java

Apex

LangGraph

LangChain

Anthropic/Claude APIs

MCP

LangSmith

Langfuse

OpenTelemetry

Promptfoo

Ragas

DeepEval

CC/CI

FLS

PII scrubbing

Claude Sonnet/Opus

Haiku

Tools

LangGraph

LangChain

Anthropic/Claude APIs

MCP

LangSmith

Langfuse

OpenTelemetry

Job description

CodeRoad provides end-to-end software development services, helping businesses scale with ideal infrastructure solutions. From staff augmentation to dedicated IT teams and general software engineering, our nearshore technology services empower businesses to thrive in an ever-evolving digital landscape.

About t he Role

As the LeadAgentic AI Engineer, you will serve as the primary technical authority responsible for architecting, engineering, and operationalizing enterprise-grade autonomous agent systems integrated directly with the Salesforce platform. You will lead the implementation of stateful multi-agent workflows, leveraging frontier LLMs (with a strong emphasis on the Anthropic Claude ecosystem), Model Context Protocol (MCP), and modern orchestration engines to bridge complex business requirements with robust software engineering.

This role is critical to transforming enterprise business processes by embedding safe, high-accuracy, and deterministic autonomous AI agents into Salesforce CRM, Data Cloud, and core operational systems. You will anchor the vision, evaluation pipelines, and architectural standards that allow high-throughput agentic workflows to operate securely within strict transactional boundaries and enterprise data security protocols.

Key Responsibilities

Architect stateful multi-agent coordination graphs using LangGraph or custom DAG loops, supporting dynamic task decomposition, supervisor orchestration, human-in-the-loop escalation, and automated error recovery.

Build standardized MCP (Model Context Protocol) servers and connectors to securely expose Salesforce tools, objects, custom actions, and operational context to agent runners.

Optimize query routes and context engineering across model tiers (such as Claude Sonnet/Opus for multi-turn reasoning vs. Haiku for high-throughput utility tasks) to optimize latency, accuracy, and token spend.

Design automated evaluation harnesses (Promptfoo, Ragas, or DeepEval) integrated into CI/CD pipelines to continuously benchmark grounding, schema validity, and hallucination resistance against reference datasets.

Implement full-stack observability and defense-in-depth security using tools like LangSmith or Langfuse alongside strict input sanitization, field-level security (FLS), and automated PII scrubbing.

Lead technical delivery standards, establishing API interface contracts, specification-driven development patterns, and architectural blueprints for cross-functional engineering teams and executive stakeholders.

Requirements

7+ years of hands-on enterprise software engineering experience using Python, TypeScript/Node.js, Java, or Apex.

3+ years of direct production experience building, evaluating, and scaling LLM-powered applications and autonomous multi-agent workflows.

Tech Stack: Expert proficiency with LangGraph / LangChain, Anthropic/Claude APIs (tool calling, prompt caching, Claude Code), MCP, and observability tools (LangSmith, Langfuse, or OpenTelemetry).

Evaluation Frameworks: Hands-on experience with automated LLM eval suites (Promptfoo, Ragas, or LLM-as-a-judge patterns) as automated test gates.

Ownership mindset with deep diagnostic debugging skills for isolating non-deterministic LLM behavior, context degradation, and latency bottlenecks.

Consultative Communication: Strong ability to translate complex AI mechanics into actionable architecture roadmaps for technical and non-technical stakeholders.

Advanced English: Fluent written and spoken professional English for direct client consultation.

Nice to Have

Practical integration experience with Salesforce APIs (REST, Composite, Bulk), SOQL/SOSL, Platform Events, and Connected Apps.

Familiarity with Salesforce Data Cloud, vector search capabilities, or Agentforce architectural patterns.

Experience deploying containerized MCP servers integrated with enterprise CRM security boundaries.

Hands-on exposure to Spec-Driven Development (SDD) methodologies for AI-assisted software delivery workflows.

• Candidate must be legally authorized to work in the United States (no sponsorship available).
• CodeRoad offers a comprehensive and competitive benefits package, including medical, dental, and vision insurance, retirement plan, and paid time off (PTO).
• The salary range for this role is $190,000–$250,000 per year.