TLDR

We are hiring Senior Forward Deployed Engineers to bring a powerful agentic AI platform into large enterprises and make it work in production.

This is a zero-to-one deployment, product-discovery, and customer-outcome role, not pre-sales or conventional implementation. You will embed with customers, solve business-critical problems, ship the smallest production-worthy solution, drive adoption, and turn field learning into reusable product.

Success means measurable value, adoption, fast time to value, safe operation, customer self-sufficiency, and reusable learning. A system nobody uses is not a successful deployment.

This role requires two qualities that rarely appear together:

Deep, hands-on ability across the full stack, cloud infrastructure, and production AI. The judgment, urgency, and presence to guide senior leaders and skeptical users through change.

If you have only one, this will not be a fit. We primarily seek mid-career to senior engineers, though exceptional early-career builders may apply.

What you will actually do

You will own enterprise AI deployments from messy problem to trusted production system: Embed with customers to understand workflows, constraints, data, systems, politics, and business pain. Diagnose the real problem, determine whether it justifies an enterprise AI deployment, and define the baseline, target, and success measures. Design, build, deploy, harden, and operate agentic AI systems. Ship the smallest credible production version quickly and iterate from evidence. Stand up multi-component services in the customer's cloud, integrated with identity, data, security, and governance. Build the CI/CD, observability, evaluation, security, and operational foundations for production. Design workflows with evals, guardrails, deterministic checks, fallbacks, human-in-the-loop controls, and monitoring. Work with frontline users, observe real work, gather feedback, and adapt until people trust and use the system. Drive adoption and navigate security, legal, procurement, data access, change control, and politics. Troubleshoot live, sometimes in front of executives, while staying composed. Teach customer teams how the system works and where its limits are. Produce documentation, runbooks, training, handoff plans, and ownership boundaries for customer independence. Feed patterns and lessons back to product and engineering, separating reusable product from customer-specific work. Contribute to core platform capabilities when a reusable pattern is clear.

What success looks like

The customer has a trusted production system that solves a real problem, creates measurable value through revenue, cost, speed, quality, reliability, risk reduction, or customer experience, is safe and supportable, can be operated independently, and teaches us something reusable.

The engineering bar

This is a senior role. You have built and operated production systems, not only prototypes or demos, and personally done most of the following:

Designed and run horizontally scaled, multi-replica services with real availability requirements. Built multi-component cloud systems using APIs, queues, caches, datastores, workers, and background jobs. Understood failure modes and improved systems after incidents. Owned multi-stage, multi-environment CI/CD with automated testing, gated promotion, rollback, and infrastructure as code. Used Docker and Kubernetes in production. Instrumented systems with metrics, logs, traces, alerts, dashboards, and incident response using Prometheus, Grafana, Datadog, OpenTelemetry, Splunk, or equivalents. Hardened infrastructure through least privilege, secrets management, image scanning, network policy, secure configuration, and supply-chain awareness. Worked within enterprise governance involving RBAC, audit, data residency, change control, and security review. Integrated APIs across messy systems and moved from infrastructure to backend to enough UI for a complete workflow. Debugged production issues under pressure. You need deep expertise in one major cloud (AWS, Azure, or GCP) and fluency in another. AWS and Azure are especially relevant, including Azure AI Foundry, Azure identity/governance, AWS VPC design, network segmentation, Amazon Bedrock, and enterprise cloud security. You must also have deployed LLM-based or agentic systems in real environments and understand orchestration, tool use/function calling, retrieval-augmented generation (RAG), evaluation-driven development, human-in-the-loop workflows, guardrails, prompt management, model tradeoffs, deterministic checks, monitoring, and safe rollout/rollback. Experience with LangGraph, CrewAI, AutoGen, Semantic Kernel, LlamaIndex, the OpenAI Agents SDK, or similar is useful. Framework choice matters less than knowing what to automate, verify, make deterministic, or keep under human control.

The customer and adoption bar

Enterprise agentic AI is as much a people problem as a technical one. Leaders may worry about reliability, security, control, cost, adoption, workforce impact, and career risk. Build confidence without overselling. You must: Build trust before building. Tell the truth about what the system can and cannot do today. Speak credibly with a CTO, CDO, CIO, CISO, VP Engineering, business sponsor, security reviewer, and frontline user. Translate between executive outcomes and engineering reality. Ask sharp questions that reveal the real workflow. Push back on the wrong approach while remaining committed to the underlying problem. Meet excited, skeptical, or anxious users where they are; teach without condescending; handle concern without defensiveness. Stay calm when things break, earn adoption through evidence, and care more about customer success than being right. The best fit combines technical depth, commercial judgment, user empathy, and founder-like ownership. The product judgment bar This role sits between customer, product, and engineering. Solve specific problems without becoming a bespoke services team.

You should be able to:

Build a fast first version without pretending it is the final architecture. Take pragmatic shortcuts without compromising security, correctness, or trust. Recognize broader patterns without generalizing too early or creating permanent customer-specific complexity. Decide what belongs in the platform, configuration, or a bespoke solution. Turn field learning into reusable capabilities. Discard or rewrite work when evidence shows it is not creating value. Technical elegance without adoption, value, or product learning is not success.

You will do well here if

You have landed difficult deployments, worked with users, and shipped production systems with real failure modes. You own outcomes end to end and accept judgment based on customer success. You move quickly with incomplete information and know what must be robust now versus improved later. You can spend days observing users or resolving access, security, procurement, and governance issues. You can connect your work to revenue, cost, speed, risk, quality, or customer experience. You value adoption and business impact over architectural elegance. You create clarity when ownership is unclear and combine persistence with humility. You can reject a requested solution, stay committed to the real problem, and discard work users do not adopt. You test new AI capabilities, judge production readiness independently, and still write, debug, deploy, and operate software.

This role is probably not for you if

You want heads-down work with little customer interaction or require clear specifications. You no longer want to write, debug, deploy, or operate software. You are a pure architect, advisor, strategist, or backend/infrastructure engineer with no interest in customer accountability. You become attached to implementations or want to perfect architecture before proving value. You see training, adoption, stakeholder management, or handoff as someone else's responsibility. You prefer theoretical correctness over shipping and evidence. You need long, uninterrupted coding periods to feel productive. You resent dependencies on security, legal, procurement, data owners, or resistant users. You prefer receiving requirements rather than discovering and challenging them. You are uncomfortable connecting work to revenue, retention, expansion, or customer outcomes. You assume autonomous agents are always better than controlled workflows. You lack strong evidence of both engineering and judgment. Regular travel or on-site work is a dealbreaker.

Before applying, ask yourself

Answer yes to most of these:

I have built and operated a production system with meaningful failure modes. I have worked directly with its users. I can describe when a customer's stated request was not the real problem. I can identify a measurable business or user outcome from my work. I have shipped quickly, learned, and substantially changed or discarded the result. I have worked through security, governance, procurement, data ownership, or organizational resistance. I have handled a production issue while the customer was watching. I have turned one customer's problem into something reusable. I am comfortable traveling, changing context frequently, and being accountable without a complete specification. I would rather solve a valuable problem imperfectly today than debate the perfect system indefinitely. If your strongest work is purely technical or purely customer-facing, this role is unlikely to fit.

Logistics Location: Remote Travel: Occasional customer travel and on-site work required Level: Senior Role Type: Full Time