Integrity Operations Manager, Sensitive Content at bumbleinc

bumbleinc

UK London·9h ago

ManagerFull-timeRustMachine LearningAgileAIML

Salary Range

£50,000 - £60,000 /year

About the Role

We are looking for an Integrity Operations Manager to join our Sensitive Content team at Bumble. This role involves owning the Human-in-the-Loop (HITL) layer of our image classification pipeline, ensuring decisions are consistent, policy-aligned, and grounded in real member impact. You will partner closely with Policy, Product, Engineering, and AI teams to continuously improve how we detect and respond to sensitive content at scale.

Key Responsibilities

→Lead day-to-day operations for the Sensitive Content pillar, ensuring accurate, timely, and policy-aligned image classification outcomes that reduce harm and protect member experience.
→Own end-to-end BPO and AI moderation vendor governance, including SLA definition, performance management, quality assurance frameworks, and structured business reviews.
→Translate sensitive content policies and taxonomy updates into clear annotation guidelines, decision trees, and workflow documentation.
→Design and evolve quality measurement frameworks, including sampling strategies, error trend analysis, reviewer accuracy tracking, and root-cause insights.
→Partner cross-functionally with Policy, Product, Engineering, and Machine Learning teams to improve moderation tooling, classifier performance feedback loops, and pipeline design.
→Coordinate special labeling initiatives, taking ownership from insight to impact with defined success metrics and clear timelines.
→Build and communicate operational reporting across quality, throughput, backlog health, escalation volumes, and cost efficiency.
→Model calm, values-led decision-making when managing high-sensitivity escalations.

Requirements

✓Typically requires 4–6 years of experience in vendor or BPO moderation operations.
✓Experience leading large-scale vendor or BPO moderation operations, including SLA management, structured QA programs, governance cadences, and distributed team performance oversight.
✓Strong working knowledge of Trust & Safety policy taxonomies and demonstrated experience operationalizing them into labeling schemas, annotation standards, and moderation workflows.
✓Hands-on experience supporting AI/ML-driven safety systems, including Human-in-the-Loop review design, dataset quality controls, calibration methodologies, and feedback loops for model improvement.
✓Comfort with operational data analysis, including building reporting dashboards, conducting trend and variance analysis, identifying error themes, and presenting insights clearly; SQL proficiency is a strong plus.
✓Demonstrated ability to collaborate with purpose across Policy, Product, Engineering, QA/Learning & Development, and external vendors.
✓Strong problem-solving judgment under ambiguity, with the ability to see things through from insight to measurable impact and adapt quickly as harm patterns evolve.
✓Thoughtful AI fluency: understanding where automation accelerates harm detection, where human judgment is essential, and how to continuously strengthen HITL systems without compromising fairness or member trust.
✓A values-driven operator who fosters psychologically safe ways of working, demonstrates Curiosity when evaluating edge cases, and upholds Respect when navigating sensitive subject matter.