Data Scientist, Trust & Safety
Job description
About the role
Replit is on a mission to create a dynamic software development platform that empowers users to build applications through natural language. In this position, you will play a crucial role in developing our Trust and Safety framework, which is essential for safeguarding the platform from malicious activities while ensuring a experience for genuine users. Your work will directly contribute to the integrity and safety of our community.
Key facts
What you'll do
- Develop a analytical framework to assess platform safety, focusing on metrics that measure abuse frequency, financial losses due to fraud, and the effectiveness of detection systems.
- dbt to create data models that consolidate signals from various sources, including payment transactions, product interactions, infrastructure usage, and identity verification processes.
- Design and implement risk assessment models and anomaly detection mechanisms to address threats such as cryptomining, phishing attempts, token farming, and exploitation of AI agents.
- Conduct shadow-mode testing and controlled experiments to evaluate the effects of new safety measures on user interactions and behaviors.
- Establish decision-making frameworks that effectively balance the need for abuse prevention with user experience and the rate of false positives.
- Analyze new and evolving abuse trends, providing actionable insights and recommendations to product and engineering teams to enhance platform safety.
- Develop monitoring systems to detect model drift and assess potential risks to legitimate users, ensuring ongoing protection against emerging threats.
- Collaborate with cross-functional teams to integrate safety measures into the product lifecycle, ensuring that Trust and Safety considerations are embedded in all stages of development.
Requirements
- A minimum of 5 years of experience in product analytics, data science, risk management, or trust and safety roles.
- Strong proficiency in Python and SQL, essential for handling extensive behavioral datasets and constructing data pipelines.
- Demonstrated experience in creating predictive models or decision-making systems, with an emphasis on navigating uncertainty and managing trade-offs.
- Ability to distill complex data insights into straightforward recommendations that can be understood by both technical and non-technical audiences.
- Experience in working with imperfect datasets, biased samples, and making high-stakes enforcement decisions.
- Familiarity with AI tools to enhance analytical processes while ensuring the maintenance of high-quality standards.
Nice to have
- Previous experience in developing anti-abuse, security, or integrity systems at a large scale.
- Hands-on experience with machine learning models, including aspects like feature engineering and optimizing thresholds.
- Knowledge in graph analysis, entity resolution, or reputation systems that can enhance Trust and Safety measures.
- Understanding of identity verification solutions such as Persona, Socure, Stripe Identity, or Prove.
- Familiarity with causal inference techniques, including synthetic control or propensity score matching.
- Experience with contemporary data stacks such as BigQuery, Snowflake, Fivetran, or Amplitude.
- Awareness of AI-related abuse tactics, including prompt injection and model extraction vulnerabilities.
Skills & tools
- Proficient in Python, SQL, dbt, BigQuery, Snowflake, Fivetran, Amplitude, Mixpanel, and Segment.
Practical notes
- The position offers a comprehensive benefits package, including a 401(k) plan with a 4% company match, health, dental, vision, and life insurance, as well as disability coverage. Additional perks include paid parental and medical leave, flexible time off, commuter benefits, and a monthly wellness stipend.
- For those working at the Foster City office, in-office amenities and setup reimbursements are provided to ensure a comfortable and productive work environment.
Join us at Replit and be part of a team that is dedicated to creating a safe and innovative platform for developers worldwide. Your expertise will help shape the future of software development, making it accessible and secure for everyone.
About the company
Replit is the agentic software creation platform that enables anyone to build applications using natural language. With millions of users worldwide, Replit is democratizing software development by removing traditional barriers to application creation.