Singapore, on-site at client premises

Six-month fixed-term contract, with extension subject to follow-on work.

Candidates must be Singapore Citizens or Permanent Residents. This role requires obtaining and maintaining government security clearance, and the work involves handling sensitive classified data on client-provided systems throughout.

About Us

Resaro was founded on the belief that AI will change the world in ways we cannot even imagine, but every new technology needs safeguards to advance.

Resaro builds custom AI testing software that helps organisations validate the performance, robustness, and safety of mission-critical AI systems, spanning computer vision, generative AI, and autonomous systems. Our clients include government, military, and commercial organisations deploying AI in high-stakes environments. We work through embedded teams deployed on-site or closely integrated with our clients.

About the Role

We are looking for an Annotation Programme Manager to run a managed data annotation programme for a client in a secure, highly regulated environment.

You are the person the client deals with. You own what has been committed on this programme: the volume delivered each month, the quality of what is delivered, and the reporting that evidences both.

You will be part of a team, and management of the annotation workforce is split between you and an annotation manager. The annotation manager runs the general annotator pool day to day. You manage the domain specialists, around 10 people whose judgement sets the quality ceiling for the programme, and you work with the annotation manager on how work is assigned across the whole team.

You also own the annotation playbook, which is what leaves the client with a repeatable capability once the engagement ends.

What You'll Do

Own the client relationship

• Act as the day-to-day point of contact for the annotation programme, including at monthly milestone reviews

• Report progress at each monthly milestone: hours delivered, quality, issues, and risks

• Raise problems early and with a proposed course of action, particularly where a delay starts on the client side

• Build the working relationship that gets the next phase of work commissioned

Manage the specialist annotators

• Directly manage a team of around 10 domain specialist annotators: deployment across tasks, performance, and development

• Train specialists on task guidelines, and re-train against specific failure modes when quality drops

• Track clearance status across your team and flag gaps that will affect the delivery plan

Run the programme

• Own the monthly delivery commitment: hours delivered against plan, across the agreed generalist and specialist mix

• Work with the annotation manager to assign work across the team as tasks arrive, and re-plan when volumes or priorities shift

• Forecast capacity against incoming task volumes and raise a shortfall before it becomes a missed milestone

• Manage attrition and keep replacement annotators moving through onboarding and clearance

• Create and implement the annotation playbook, adapting it based on what pilot and iterative trial runs show

Own quality

• Define, apply, and document the quality criteria and verification thresholds for each task type

• Decide how much of the output gets verified, and concentrate that effort where errors actually appear

• Work out why quality drops when it does: unclear guidelines, weak machine pre-annotations, fatigue, or genuinely hard data are different problems with different fixes

• Trigger retraining when a task type or an annotator falls below threshold

What We're Looking For

• 5+ years managing operational delivery, including direct line management of a specialist or technical team

• Demonstrated ownership of a client-facing delivery commitment, where you carried the volume and quality numbers yourself

• Experience in data annotation, data labelling, transcription, QA, or comparable structured-production work, ideally including video or image tasks

• Comfortable managing from data: throughput per annotator, hours against plan, pass rates by task type

• Strong written English and a track record of producing SOPs, playbooks, or operating manuals that other people successfully used

• Able to share responsibility for a workforce with a peer manager rather than holding the whole reporting line

• Willing and able to work on-site in a secure facility, on client-provided hardware, under classified data-handling rules and statutory confidentiality undertakings

• Self-directed. Task assignments arrive on a rolling basis and the process is still being proven during the engagement

Nice to Have

• Familiarity with annotation quality metrics, including inter-annotator agreement, and with using them to decide where to intervene. Our scientists can support you on the analysis, so this is helpful rather than essential

• Prior work inside a Singapore government or other secure, cleared environment, and familiarity with clearance timelines

• Familiarity with annotation tooling and with model-assisted labelling workflows

• Exposure to computer vision or vision-language model training and evaluation, enough to understand what the downstream team needs from the data

• Experience standing up an annotation function rather than only operating an existing one

• Basic data handling skills for reporting: spreadsheets to a competent level, SQL or Python a bonus

Our Hiring Bar

• Communication and handover: You must be able to clearly document your work, hand over context to colleagues, and communicate operational decisions to both annotators and client stakeholders.

• Independent operation: You will be embedded on client premises with limited daily supervision. You must manage ambiguity and make sound calls alone.

• Sound judgment under incomplete information: You must surface risk early, escalate with a proposed course of action rather than a problem statement, and hold a defensible position with the client.

• Ownership: You own the delivery numbers for this programme.

Engagement Terms

• Duration: Six months from contract start, against monthly delivery milestones

• Commitment: 125 hours per month, roughly 15 to 16 working days, on-site at client premises as required for system access

• Reporting line: Resaro engagement lead

• Compensation: Competitive day rate, commensurate with experience

• Conditions: Appointment is subject to client approval. Signature of statutory confidentiality undertakings is required before work begins.