Data Scientist, Trust & Safety

You're a data scientist who moves fast, goes deep, and thinks adversarially. You can spin up an analysis in hours that would take others days, not by cutting corners, but because you've built the intuition and technical toolkit to get to the right answer quickly. You dig past the top-line abuse rate to understand selection effects, missing labels, policy changes, attacker adaptation, and the false positives hidden inside an aggregate metric.

You understand that Trust & Safety data is imperfect and outcomes are high stakes. Ground truth is delayed, biased, and often incomplete; attackers react to defenses; and an apparently effective rule can quietly harm legitimate users. You pressure-test your own work, quantify uncertainty, and distinguish correlation from evidence strong enough to justify enforcement.

You use AI agents and tools aggressively to multiply your output: writing code, exploring data, generating hypotheses, and prototyping investigations. But you treat every AI-assisted output as a draft, not a deliverable. You know what good analysis looks like and won't ship anything that doesn't meet that bar.

You Will

• Own the analytical foundation for Trust & Safety, including abuse prevalence, fraud loss, false-positive and false-negative rates, time to detect, time to mitigate, appeal and reversal rates, and verification step-up conversion.
• Build reliable datasets and dbt models that connect product events, account and identity signals, payment activity, infrastructure usage, content classifications, enforcement actions, appeals, and support outcomes.
• Develop and evaluate risk models, rules, and anomaly-detection systems for threats such as phishing, scam hosting, cryptomining, token farming, payment fraud, promotional abuse, and AI-agent exploitation.
• Design rigorous offline evaluations, shadow-mode tests, holdouts, and controlled experiments to measure detection quality and the user impact of new policies, enforcement actions, and progressive verification.
• Define thresholds and decision frameworks that balance abuse reduction, economic loss, customer friction, and false positives across free, paid, and enterprise users.
• Investigate emerging abuse patterns, quantify their impact, identify coordinated behavior, and turn ambiguous signals into clear recommendations for product and engineering teams.
• Develop predictive models that estimate account, device, transaction, workspace, or deployment risk and embed those signals into detection, review, and escalation workflows.
• Partner with Support and Legal to improve case review, appeals, reason-code quality, and feedback loops so human decisions become useful model and policy signals.
• Build monitoring that detects model drift, attacker adaptation, data-quality failures, and unexpected harm to legitimate users.
• Communicate findings clearly to technical and non-technical partners, including the tradeoffs, uncertainty, and evidence behind high-impact decisions.

Examples of What You Could Do

• Build a measurement framework for Replit's abuse surface, reconcile incomplete labels across automated detections, human review, appeals, chargebacks, and support cases, and establish a trustworthy baseline for the first time.
• Design and evaluate a risk-scoring model for suspicious account clusters using identity, device, payment, graph, and product-behavior signals, then define thresholds that materially reduce fraud while protecting legitimate users.
• Analyze a phishing detection rule that appears highly precise, uncover that it disproportionately bans paying users with legitimate brand references, and redesign its evaluation and review path to reduce false positives.
• Measure a progressive verification "ladder of trust," determining when to step users up to additional verification and quantifying the tradeoff between abuse prevented and legitimate-user conversion lost.
• Detect coordinated token-farming or promotional-abuse networks by combining account-linkage graphs, referral behavior, payment patterns, and infrastructure usage, then partner with Engineering to operationalize the findings.
• Evaluate a new enforcement policy in shadow mode, estimate its counterfactual impact, and recommend whether to launch, revise, or reject it before any users are affected.
• Experience building or evaluating anti-abuse, fraud, identity, security, spam, integrity, or content-safety systems at scale.
• Built, shipped, and maintained ML models in production (classification, anomaly detection, or risk scoring), including feature engineering on behavioral and transaction data, threshold selection against precision/recall economics, and post-launch monitoring
• Experience with graph analysis, entity resolution, coordinated-behavior detection, reputation systems, anomaly detection, or risk scoring.
• Experience measuring false positives and enforcement harm, designing human-review workflows, or using appeals and case outcomes as model feedback.
• Familiarity with progressive verification, KYC, account trust, or identity providers such as Prove, Persona, Socure, or Stripe Identity.
• Experience with causal inference methods such as difference-in-differences, propensity score methods, synthetic control, or uplift modeling.
• Experience with a modern data stack such as dbt, BigQuery, Snowflake, Fivetran, Amplitude, Mixpanel, or Segment.
• Experience at a consumer platform, developer tool, cloud provider, marketplace, fintech company, or other product with a meaningful adversarial surface.

Bonus Points

• You've built AI-powered analytical tools, investigation systems, automated detections, or novel measurement approaches.
• You have experience with AI-native abuse such as prompt injection, LLM token farming, model extraction, or agent-driven abuse.
• You understand freemium, usage-based, or promotional pricing models and the abuse incentives they create.
• You've worked directly with operational review teams and can translate analytical signals into practical playbooks, queues, and escalation paths.

This is a full-time role that can be held from our Foster City, CA office. The role has an in-office requirement of Monday, Wednesday, and Friday.

• 5+ years of experience in data science, product analytics, fraud, risk, trust and safety, or a related field.
• Strong SQL and Python skills, with experience working with large behavioral datasets and building reliable data models or pipelines.
• Experience developing and evaluating predictive models, experiments, or decision systems, with sound judgment around uncertainty and tradeoffs.
• Ability to turn ambiguous data into clear recommendations and communicate them effectively across technical and non-technical teams.
• Comfort working with imperfect labels, biased samples, and high-impact decisions where false positives matter.
• You use AI tools extensively to increase your effectiveness while maintaining a high bar for analytical quality.