Forward Deployed Engineer

About The Thinking Company

The Thinking Company (TTC) is an AI transformation firm. We work with CEOs, boards, leadership teams, and private-equity operating partners when AI needs to change the economics of a business.

Clients buy defined work across AI transformation, AI product and build, and private-equity value creation and post-M&A transformation. We identify where value can be created, deploy a system or operating change, measure the result, and transfer ownership to the client.

Runstate is the agentic system behind TTC. It combines our delivery methods, engagement context, and AI agents to help teams move from technical discovery to production and handover. FDEs use it every day alongside Transformation Partners and client teams.

Why this role is different

• Own ambiguity through to a reliable production system the client can operate.
• Work directly with users and senior stakeholders, accountable beyond the prototype.
• Work across architecture, production code, AI evaluation, deployment, stabilization, and handover.
• Build the smallest valuable system; use AI only when it is the right engineering choice.
• Use and improve Runstate; earn broader deployment authority through judgment and evidence.

The mandate and measures of success

Your mandate is to turn an ambiguous client problem into a reliable production system the client can operate, from discovery and architecture through stabilization and handover.

A client may ask for a tool, model, or pilot. You will decide what should be built and define the smallest robust system that creates the agreed value. The stack varies, demanding broad production judgment and fast learning.

At steady state, you will own one intensive deployment and one lighter engagement in discovery, support, or handover. You will not carry two critical go-lives at once.

What you will own

• Turn an incomplete brief into a clear user problem, technical scope, acceptance criteria, risk boundary, and release plan.
• Choose the architecture, models, data flows, integrations, and build-versus-buy path needed for the result.
• Write and materially review production code across the stack required by the engagement.
• Build model-driven workflows, agents, retrieval, tool use, evaluations, and human controls when they fit the problem. Use conventional software when it is the better choice.
• Own testing, evaluation evidence, observability, release, recovery, stabilization, and the record of accepted limitations.
• Diagnose production issues across TTC's work, client systems, and third-party services. Fix the cause where the scope allows it.
• Work directly with client engineers, operators, security teams, and users. Keep the people who will run the system involved in the decisions.
• Plan handover early. Leave the code, configuration, evaluation records, operating procedures, and known limitations in the client's approved systems of record.
• Verify that the client team can deploy, monitor, diagnose, recover, and make an agreed change before TTC steps back.
• Assess the next technical opportunity at the end of each engagement. Recommend further work only when the value, feasibility, data, and operating conditions support it.
• Contribute only TTC-reviewed, generalized learning to shared methods. Keep client-specific code, data, and material inside approved client systems.

Working with Runstate, Transformation Partners, and AI agents

You will use Runstate every day to structure technical discovery, maintain system and decision state, coordinate work with AI agents, build evaluation evidence, record technical decisions, and prepare releases and handovers.

You will work with a Transformation Partner and client teams as one delivery unit. The Partner owns the institutional result; you own technical delivery. The people defining the system stay involved through deployment.

You will improve Runstate with reusable engineering patterns. Feedback is direct, decisions are recorded, and quality is demonstrated through working systems, operational evidence, and successful transfer.

What the role requires

The preferred candidate is a Forward Deployed Engineer or software engineer who has shipped systems used by real people and owned them after release. You can move from user problem to production behavior and work across code, APIs, data, infrastructure, and operations.

Production AI experience is strongly preferred. An exceptional production engineer can qualify through substantial current AI work, an excellent practical demonstration, and clear evidence of fast learning.

You should have a current working view of model behavior, agent patterns, evaluation, and the controls required when probabilistic systems touch real work. Agentic engineering is part of your daily practice. You can explain where it saves time, where it adds risk, and how you verify the resulting code.

The team is small. There is no guaranteed architecture, QA, solutions, or DevOps layer to absorb the edges of an engagement. The workload becomes demanding around go-lives, and production discipline still applies under time pressure.

Automatic disqualifiers

• You wait for a complete specification instead of shaping the problem.
• Your work stops at a demo and leaves testing or operations to someone else.
• You avoid direct work with clients, users, training, or handover.
• You build a platform before the first deployment has proved the need.
• Your AI experience is commentary and experimentation without a serious engineering practice.
• You can explain the system but cannot personally ship it.

Limited research credentials, uneven visual-design polish, and limited people-management experience are acceptable. Gaps in enterprise systems, executive communication, or stack breadth require strong evidence that you can learn the missing area quickly.

Evidence we want to see

We care about systems you owned and the decisions behind them. Be ready to discuss:

• A production system you shaped, shipped, and operated after release.
• A failure or incident you owned, including the permanent change that followed.
• A client or user request you reframed before building.
• A decision to build a smaller system and what that choice enabled.
• An AI workflow or agent, including its evaluations, controls, and known failure modes.
• A system another team operated or extended after your handover.
• Your current coding-agent workflow and the checks that keep generated code out of production until it is understood and verified.
• A technical opportunity you rejected because the value or operating conditions were weak.

We assess the examples above more heavily than employer names or framework keywords.

Location, travel, and language

Warsaw-first is preferred. A Poland-based hybrid arrangement is possible for an exceptional candidate.

The role includes client-site work across Europe from time to time for discovery, build, go-live, and handover. Professional English is required for client work. Polish is helpful but not required.

Apply

Send your CV or LinkedIn profile, a short note explaining why the role fits, and at least one example of relevant work you personally owned.

Apply for Forward Deployed Engineer — [email protected]