Forward Deployed Engineer (AI)

- Compensation: €38,000 to €47,000 gross annual salary, plus an annual benefits budget of €15,000 to €19,000, including meal allowance. Total annual package: €53,000 to €66,000. Profit sharing: in addition to salary and benefits, we share company profits with the team, equivalent to around 8% of annual gross salary on average. Final payouts depend on company results and the scheme’s allocation rules.

- Travel: Occasional client travel, typically for discovery, building sessions and important rollouts

Whitesmith is an AI-first technical partner with deep roots in senior software engineering. We help established organisations make AI part of how their teams actually work: not a slide deck or an isolated proof of concept, but working systems with measurable value and people who can keep improving them.

We are hiring a Forward Deployed Engineer who wants to learn how to solve hard problems together with users, client engineers and our Engagement Lead. You do not need every answer upfront; you do need to own the next responsible step and the system you build.

You will work directly with users and client engineers across multiple workstreams and clients alongside a Whitesmith AI Engagement Lead. Allocation will reflect planned delivery days, technical complexity, support commitments and the cost of switching context. You will take a shaped but still ambiguous problem from workflow discovery through a useful prototype, production or regular operational use, and the iteration loop that keeps it working.

This is not a demo-engineering, support or ticket-delivery role. You will write production code, integrate with real systems, define evaluations and guardrails, debug failures and help people adopt what you build.

As the primary technical point of contact for your assigned workstreams, you will earn the trust of CTOs, technical stakeholders and engineering teams through engineering depth, delivery speed and sound technical judgment. You need the curiosity to understand work that is not yours yet and the ability to explain your technical decisions clearly.

What you will own

The working system. Build the smallest useful version that runs the whole workflow, then harden what earns the right to stay. Work across application code, APIs, data, automation and agentic workflows as the problem requires.

Technical discovery. Sit with users, inspect the existing workflow and find the constraints hidden by the first brief. Turn them into a technical plan, acceptance criteria and a fast learning sequence.

AI quality. Define what good looks like. Build evaluations, tests, observability, fallbacks and human controls appropriate to the consequence of failure. Treat fluent output as unverified output.

Integration and production. Work within the client’s identity, security, data and infrastructure constraints. Make deliberate trade-offs across quality, latency, cost, maintainability and speed.

Adoption and handoff. Watch how the system behaves in real work. Train the team, write the runbook and make the client progressively less dependent on you.

The reusable pattern. Turn useful field learning into Whitesmith skills, knowledge, evaluation sets, templates and building blocks. Resist one-off complexity when a simpler or reusable path exists.

What good looks like after six months

- You can own the technical delivery of a client workstream from a messy workflow to stable regular use.

- The work has a measurable outcome and an evaluation or feedback loop, not just a successful demo.

- Users and client engineers trust you with both the system and the honest limits of the technology.

- Security, data access, failure handling, observability and support are dealt with before they become production surprises.

- The client team can operate and improve what you built with clear documentation and training.

- At least one pattern from your account has become reusable Whitesmith IP.

- Your own AI-enabled engineering system makes you materially faster without reducing quality or understanding.

You may be a fit if

- You have shipped production software to real users and can explain your personal contribution from architecture through operation.

- You have built useful systems with LLMs or agents and understand prompting/context, tool use, retrieval where appropriate, evaluation and common failure modes.

- You can move from an unfamiliar business workflow to a working vertical slice without waiting for a perfect specification.

- You use AI deeply in your own engineering work. You can show where it accelerated you, where it failed and how you verified the result.

- You communicate clearly with users, engineers and leaders. You can say “this will not work as proposed” and bring a better path.

- You enjoy learning with others through unfamiliar problems, seek useful feedback and keep improving the work after the novelty wears off.

Consulting, solutions engineering, implementation or early-stage product experience can help; none is mandatory. Sound production engineering and sophisticated AI use are essential. Beyond that bar, we value motivation, follow-through and learning with others over pedigree or quick answers.

This role is probably not for you if

- You enjoy prototypes but lose interest in integration, testing, rollout and support.

- You want product requirements to arrive complete and stable.

- You use AI to produce code you cannot explain, test or debug.

- You optimise for technical novelty before user value.You would rather receive tickets than spend time with the people doing the work.

- You treat documentation, training, privacy or security as cleanup for somebody else.

How we hire and how AI fits

AI is part of the job, so it is part of the technical assessment. You will use an AI coding assistant in a realistic, synthetic repository while we observe how you frame the work, inspect the codebase, direct the tool, test the result and recover from mistakes. You will then explain and modify part of the work without AI so we can establish the understanding underneath the leverage.

We care about deliberate AI use and accountability, not a memorised interface. Expect to attempt a problem, receive feedback and improve the solution; asking for specific help is welcome, while ownership stays with you.

A standard environment is available without a personal account. We will not ask for unpaid client work.

Our six stages are: a five-minute demo; short recorded answers; a live structured screen; a practical exercise with reference checks in parallel; an in-person discovery and team session; and a final decision.

We aim to respond within two working days after each stage and complete the process in about three to four weeks. If delayed, we will explain why and give a revised decision date.

Whitesmith is an equal-opportunity employer. We welcome applicants with different backgrounds and will provide reasonable adjustments throughout the process. If an assessment constraint conflicts with an accessibility need, tell us and we will agree an equivalent way to demonstrate the skill.

Where's your next real AI gain? A short check reads where your organisation's AI momentum is building, where it's stalling, and the moves that matter most next.

#J-18808-Ljbffr