Forward Deployment Engineer

Forward Deployed Engineer About the role You earn trust by solving the customer’s problem with working agentic AI systems. The work starts close to the user and ends close to production. Location : Hybrid (Dallas, Santa Clara, New Jersey) In this role you design, build, integrate and deploy agentic AI solutions inside customer environments. You turn unclear needs into working systems: AI agents, multi agent workflows, tool and API integrations, retrieval and data pipelines, evaluations and the cloud-native platforms that run them. You own this end to end, from discovery and design through build, integration, and production rollout, working alongside the customer’s engineering and domain teams. You work across the major agentic AI stacks rather than a single vendor. Depending on the engagement, that means building with Anthropic, OpenAI, Google or Microsoft, and with vendor-neutral frameworks such as LangGraph where the customer wants to stay portable across models. You prove the idea quickly in a visual builder when that helps the customer. You then move to governed, code-first delivery, where evidence, tests, evaluation, and release confidence turn the prototype into verified value. You pick the right model, framework, and harness for the customer’s constraints, then make it reliable in their environment. Success is measured by what the customer can run in production and the value it delivers, not by the demo. In the FDE Practice this role covers both value engineering and quality engineering. The value engineering emphasis is implementation and user value. The quality engineering emphasis is evaluation, reliability, release safeguards, and production readiness. You may lean one way or the other depending on the engagement. You will use coding agents where they increase speed, but you stay responsible for direction, review, and control. You supervise agent output, inspect assumptions, write or refine code and hold the line on enterprise delivery standards. Location and level • Location: Hybrid, with client-site travel where co-build, discovery or rollout requires it. • Level: Senior. This is a hands-on role for an experienced engineer who can own delivery in front of a customer. What you will do • Build agentic AI solutions: single and multi-agent systems, tool use, retrieval pipelines, automation, and prototypes that reach production. • Build on the major agentic stacks, including the Claude Agent SDK, OpenAI Agents SDK, Google Agent Development Kit, Microsoft Agent Framework, and vendor neutral frameworks such as LangGraph, choosing the right one for the customer’s needs. • Work with customer technical teams to understand systems, constraints, data, deployment paths, and operating needs. • Use coding agents and human engineering judgement to move quickly while preserving review, test, and release discipline. • Integrate agents with enterprise systems, APIs, cloud services, databases, ticketing systems, model services, and security controls, often through MCP and similar tooling. • Implement agent evaluations, automated tests, observability hooks, runbooks, and release safeguards. • Troubleshoot implementation, deployment, data, model, and integration issues. • Codify what works into reusable patterns, starter kits and playbooks, alongside technical decisions, defects, risks, and lessons learned, to raise the floor for the wider practice. • Feed field insight back to the platform vendors through our partnerships, turning recurring friction and gaps into product feedback that shapes the tools you build on. • Support handover and production readiness with clear documentation and operational ownership. • Work with architects and delivery managers to keep implementation aligned to value and technical guardrails. What we are looking for These are the essentials. If you meet most of them, we want to hear from you. • Several years of production software engineering experience in Python, TypeScript, Java, or comparable stacks. • Hands-on experience building with LLMs and agents on at least one major platform, such as Anthropic, OpenAI, Google or Microsoft. • Practical understanding of agentic patterns: tool use, retrieval, prompting, model behavior, evaluation and multi-agent workflows such as orchestration, delegation, and self-reflection. • Experience with APIs, databases, cloud platforms, integration surfaces, or comparable deployment environments. • Strong debugging, testing, and integration discipline. • Experience working in ambiguous delivery contexts with customers, users, or cross functional technical teams. • Ability to balance fast prototyping with production-quality engineering. Nice to have These would strengthen your application, but they are not deal-breakers. We do not expect every candidate to bring all of them. • Experience taking agentic or AI systems from prototype to production rollout. • Experience with agent orchestration frameworks, MCP servers or custom tool integrations. • Experience with data and retrieval pipelines, workflow automation, or model deployment. • Experience with evaluation, observability, troubleshooting, or release safeguards for AI systems. • A vendor certification on one of the major AI platforms (Anthropic, OpenAI, Google or Microsoft). Who does well here The strongest people in this role are: • Hands-on and pragmatic, with a bias towards working software. • Careful about quality, testing, integration, and operational consequences. • Comfortable pairing with customers, architects, engineers, and agents. • Quick to learn unfamiliar domains, APIs, and systems. • Clear in writing, especially around decisions, assumptions, tests, and handover notes. • Willing to challenge generated code and weak assumptions rather than accept output at face value. • Focused on user value without ignoring maintainability and production risk. Our process We keep the process short and transparent. There are three stages, and we give you feedback at each step. 1. First conversation (about 45 minutes). A conversation with the hiring manager or a senior engineer about your background, motivation, and the breadth of your experience across software, AI and client delivery. We also use it to explain the role, the pod model and how the practice works. 2. Technical interview (about 60 minutes). A deeper session on technical depth and communication. You present a previous project that fits the role well, then talk through a short delivery scenario, such as how you would frame and deliver an agentic proof of concept. We send the brief 48 hours ahead so you can prepare. We are not looking for polish or for you to pick the same tools we would; we want to see how you think. 3. Final conversation (about 60 minutes). A conversation with the practice lead about how you work, how you learn and whether the environment is right for you. This is a discussion, not a technical grilling. In one line This is a hands-on field engineering role for someone who can build agentic AI systems beside the customer. You make the work useful, reliable, and ready for real delivery. Show more Show less