AI Engineer

Insight Global • Dunwoody, GA

6d

About The Position

Design and build agentic workflows—multi-step, multi-agent systems that reason, use tools, call external services, and recover from failure—and take them from prototype to reliable production. Translate business requirements into working systems by composing agents, designing prompts and behaviors, connecting tools and data sources, choosing coordination approaches, and deciding where human review is needed. Build on the company’s agent development tooling, which provides identity, guardrails, logging, and monitoring, while focusing on assembling those capabilities into predictable solutions. The role is hands-on and product-facing: translate requirements into designs, work with real data, harden systems for production, and improve them after launch.

Work with product and engineering partners to build practical, operable, and governable agent systems. Contribute reusable patterns and components that improve how agentic systems are built across the team.

Requirements

• Strong software engineering fundamentals and production-grade Python skills, including clean, tested code, async patterns, and long-running stateful processes.
• Experience building or working with agentic or LLM-driven systems in production or near-production settings, where models drive control flow through tool use, multi-step reasoning, or agent handoffs.
• Experience composing multi-step or multi-agent workflows, orchestrating agents, managing shared state, chaining tool calls, and handling failures and retries.
• Practical prompt-engineering skills, including designing prompts and system messages for consistent behavior and improving them through evaluations.
• Experience integrating external APIs and services, including REST, authentication, error handling, and designing for partial failure.
• Experience building tools or integrations that models or agents call, including function interfaces, schema design, or service endpoints.
• Ability to describe an evaluation suite built for an agentic or LLM system, how ground truth was created, and what the evaluation demonstrated.
• Strong debugging skills for distributed and non-deterministic systems involving failures across agents or services.
• Ability to write meaningful unit, integration, and scenario-level tests for complex stateful systems.
• Familiarity with agent or LLM observability, including tracing, logging, or monitoring model and tool calls.
• Backend engineering background, including APIs, queues, and databases.
• Exposure to multi-tenant systems and data-isolation concerns.

Nice To Haves

• Experience with durable or long-running workflow execution where state must survive restarts, retries, or partial failures.
• Familiarity with agent memory, including working memory, long-term memory, and context-window management.
• Experience with retrieval-augmented generation and reranking, including integrating retrieval into agent workflows and measuring its effect.
• Exposure to human-in-the-loop patterns, including production approval steps, escalation policies, and review workflows.
• Experience with agent orchestration frameworks or patterns such as LangGraph or comparable approaches.
• Familiarity with prompt-injection and adversarial risks in agent and retrieval systems and methods to mitigate them.
• Comfort working with an SDK or framework abstraction, understanding its constraints, and knowing when to escalate gaps.
• Open-source contributions to agent frameworks or related tooling.

Responsibilities

• Design and build agentic workflows, owning them from design through production.
• Compose agents into reliable pipelines integrating external APIs, internal services, and data sources, with clear error handling and retry semantics.
• Design and iterate on prompts and agent behaviors as versioned, tested artifacts, validating them against evaluation sets.
• Build safe and correct tools and integrations, such as typed function interfaces or service endpoints, for agents to use.
• Choose appropriate agent coordination approaches, such as supervisor-worker, planner-executor, critique, or fixed-sequence patterns.
• Determine where human review and approval are needed and build checkpoints into long-running workflows.
• Configure state and memory for agent workflows while preventing data leakage across users or tenants.
• Build evaluation suites and regression checks that assess trajectory and outcome quality.
• Harden production workflows by diagnosing timeouts, partial failures, unexpected model behavior, and state drift.
• Instrument and debug agent behavior using tracing, logs, and cost and latency telemetry.
• Contribute shared patterns, conventions, and reusable components for building agentic systems.
• Partner with product and engineering teams to translate business requirements into practical, operable, and governable agent designs.

What This Job Offers

• Job Type: Full-time
• Education Level: No Education Listed