Data Scientist, Trust & Safety

Replit•Foster City, CA

2d•$210,000 - $310,000•Onsite

About The Position

Replit is seeking a Data Scientist to help build its Trust & Safety and Anti-Abuse program. This role involves transforming various signals (behavioral, identity, payment, infrastructure, content) into measurement systems, detections, and decisions to protect users, the platform, and economics. The Data Scientist will collaborate with Engineering, Support, Legal, Security, Infrastructure, Money, and Growth teams to make abuse economically unviable while minimizing friction for legitimate users. The role is crucial for addressing AI-native abuse, including phishing, scam hosting, cryptomining, LLM token farming, fraud, referral abuse, and AI agent-driven abuse, by defining methods to identify, measure, and respond to these threats without compromising user experience.

Requirements

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.

Nice To Haves

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.
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.

Responsibilities

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.