Director of Operations

Build the operation that makes the service scale

Remoat Teams is hiring an AI-native operator to build and scale technology-enabled services. The role could sit in recruitment operations or AI data labelling, depending on your strengths and the business priority. We need someone who can turn a client requirement into a working delivery system and improve that system as demand grows.

The operating challenge is concrete. A client needs a reliable outcome by an agreed date. In recruitment, that means qualified people progressing through a hiring process. In AI data labelling, it means a capable workforce producing work that meets a defined quality standard. Both require dependable supply, clear instructions, sensible allocation, quality checks and early visibility of delivery risk. You will own how those pieces work together.

If you have built a marketplace, launched a service or scaled an early-stage operation, you will recognise the problem. Demand is uncertain, capacity takes time to build, quality cannot slip, and adding people to every broken process is not a scalable answer. The opportunity is to build a repeatable operation in one of these domains, not to pretend they have identical workflows or quality requirements.

Read this first

This is a demanding startup role for someone who wants real operating responsibility. You will work close to the founder and the delivery, recruiting, product and engineering teams. Expect incomplete information, priorities that change and decisions that cannot wait for a perfect plan. Being AI-native is a requirement throughout the work, not an optional skill added to the job description.

Weekend work is normal here, not an occasional exception. If you want a predictable Monday-to-Friday job, a narrow task list or a role where someone else makes every difficult decision, do not apply. We expect high urgency, direct communication and work that holds up in practice.

We will discuss availability, time-zone overlap and the working arrangement during the hiring process. Exact hours, working days and rest arrangements will be confirmed before an offer. We want a clear understanding of the commitment, not surprises after you join.

What you will build and run

• Client launch plans. Turn a confirmed brief into milestones, owners, capacity requirements and acceptance criteria. Establish what needs to be ready before a hiring ramp or data-labelling project goes live. Identify what can run manually at first and which dependencies need resolving now.

• A reliable supply engine. Build the approach to attracting, qualifying and activating the people needed to deliver. In recruitment, model the funnel through to qualified starts. In data labelling, model contributor availability, task eligibility and productive capacity. Test sourcing and engagement changes against real outcomes rather than raw application counts.

• Workflows that hold together. Map qualification, assessment, onboarding, work allocation and handoffs. Recruitment needs defensible candidate decisions. Data labelling needs clear task guidance, quality review and a route for uncertain or disputed labels. Work with domain experts to establish the right checks instead of imposing one generic process on both.

• Useful operating metrics. Build a reliable view of supply, conversion, queue time, throughput, quality, rework and cost. Use the measures appropriate to the assignment, such as qualified hiring starts or accepted labelled output. Use AI to investigate patterns and prepare analysis, then validate the data and make the operating decision.

• AI-native execution. Use LLMs and agentic tools to analyse bottlenecks, develop working procedures, build lightweight automations and test alternatives. You must understand how to use tools such as Claude Code, Codex, GrokBot and Dots within ChatGPT, and choose the right approach for the task. Show how you validate outputs, respect data-access permissions, handle exceptions and keep human judgment where it matters.

• A productive product partnership. Bring product and engineering a clear problem, real examples, operational requirements and acceptance criteria. Prototype an appropriate lightweight solution with AI tools when you can, without bypassing security or production controls. Distinguish an urgent workaround from a reusable improvement and test the result with its users.

• Repeatable delivery across clients. Turn successful experiments into playbooks, reusable prompts, workflows and clear ownership. Keep the parts that should be standard and adapt what genuinely differs by client, task, language or quality requirement. The next launch should benefit from the last one.

• Independent decisions and early escalation. Run a clear operating cadence, keep client-facing colleagues informed and close issues through to resolution. Make calls within your remit. When a decision needs a founder or commercial owner, bring the options, consequence and recommendation.

The kinds of problems you will solve

A recruitment channel produces applications but few qualified starts. Or a labelling queue processes quickly but fails quality review and creates rework. You will investigate the actual constraint, use AI-assisted analysis to test explanations and decide whether to change supply, instructions, screening, allocation or quality control.

A client launch depends on a workflow that engineering cannot finish in time. You will define the smallest safe process that can work now, use approved AI tools to prototype or automate suitable steps, make the manual cost visible and specify what needs to be built next.

Demand changes after a plan is agreed. You will update the supply and capacity assumptions, identify the trade-off and communicate what is achievable. You will not quietly preserve an unrealistic deadline or compensate by adding unmeasured activity.

These are examples of the judgment we need, not claims about a particular client or a fixed project assignment.

Why this is worth your time

• Build while choices are open. Shape how a service operates, how people and AI work together, how quality is protected and how the operation learns. Your contribution is a working system, not another layer of reporting.

• See the effect of your decisions. Better execution should show up in launch readiness, qualified starts or accepted output, reliable quality and the cost of delivery. You will be close enough to the work to learn whether a change actually helped.

• Operate across disciplines. Work across delivery, product and engineering rather than optimise a small part of a mature process. Build AI-native working habits into the operation from the beginning.

We are looking for the operating habits associated with strong startup teams. fast learning, sound judgment, ownership and useful output. You do not need to have worked in San Francisco or at a famous company. You do need to show that you can build.

Evidence that will get our attention

• Proven operating ownership. You have personally launched or substantially improved a startup, marketplace or technology-enabled service. Explain the starting constraint, your own decisions and the result.

• Quantitative judgment. You understand how demand connects to supply, productive capacity and service quality. Build a simple model, explain its assumptions and revise it when reality disagrees.

• AI-native fluency. You use modern LLM tools as a core part of how you work, not only to draft text. You can demonstrate practical use of Claude Code, Codex, GrokBot or Dots within ChatGPT, explain how you choose and combine tools, and transfer that understanding across the toolset. You must validate outputs, understand limitations and work within approved data and access boundaries. You do not need to be a software engineer, but you must be able to build useful workflows.

• Hands-on systems thinking. Inspect a broken handoff yourself, then improve the workflow so the team does not need you at every handoff. Use AI to increase leverage without hiding weak process or poor data.

• Clear decisions and feedback. Communicate a recommendation, explain the trade-off and respond constructively to direct feedback. Defend a supported view and change your mind when the evidence changes.

Relevant backgrounds include marketplace operations, startup operations, service launches, strategy and operations, AI data operations, or a hands-on Chief of Staff role. Recruitment or data-labelling experience helps, but neither is required for this broader builder profile. We will consider strong Manager or Senior Manager candidates whose scope exceeds their title.

What success should look like

Early on, you should explain the live priorities, establish a credible supply and capacity view, own an agreed bottleneck and put an improved workflow into use. Success means more predictable launches, stronger qualified-candidate conversion or accepted-output throughput, fewer quality failures and less repeated manual work. The relevant measures depend on the operating domain you take on.

We will agree specific milestones against the actual client brief, baseline and available resources. This role owns operating execution and informs product priorities; final product direction, pricing and contract commitments remain with the relevant business owners.

Do not apply if

• Strategy without execution. You want to advise without responsibility for implementation and results.

• Dependence on instructions. You need a finished playbook or detailed tasks before taking useful action.

• Credentials instead of evidence. You rely on presentation quality, employer names or a senior title rather than explaining the work you personally did.

• A predictable working week. You are uncomfortable with normal weekend work, changing priorities or candid feedback.

• AI as an optional extra. You do not use LLM tools in real work, are unwilling to become fluent across the named toolset, or trust generated outputs without checking them.

How to apply

Send your CV and one concise example of an operation you built or improved. Explain the starting problem, your own decisions, the measurable result and what you would do differently. An anonymised example is welcome; do not share confidential information from another employer or client.

Our proposed selection process includes an introductory conversation, a short practical exercise, a team discussion and a final conversation. We will share the exercise scope and time expectation in advance. Disclosed AI use is welcome; we will assess your judgment and ability to explain and adapt the output.