Machine Learning Scientist - Trust Detection

Depop is the community-powered circular fashion marketplace where anyone can buy, sell and discover desirable secondhand fashion. With a community of over 35 million users, Depop is on a mission to make fashion circular, redefining fashion consumption. Founded in 2011, the company is headquartered in London, with offices in New York and Manchester, and in 2021 became a wholly-owned subsidiary of Etsy.

Find out more at

Our mission is to make fashion circular and to create an inclusive environment where everyone is welcome, no matter who they are or where they’re from. Just as our platform connects people globally, we believe our workplace should reflect the diversity of the communities we serve. We thrive on the power of different perspectives and experiences, knowing they drive innovation and bring us closer to our users.

We’re proud to be an equal opportunity employer, providing employment opportunities without regard to age, ethnicity, religion or belief, gender identity, sex, sexual orientation, disability, pregnancy or maternity, marriage and civil partnership, or any other protected status. We’re continuously evolving our recruitment processes to ensure fairness and are open to accommodating any needs you might have.

If, due to a disability, you need adjustments to complete the application, please let us know by sending an email with your name, the role to which you would like to apply, and the type of support you need to complete the application to  For any other non-disability related questions, please reach out to our Talent Partners.

At Depop, machine learning is integral to building a safe and trusted marketplace. As a Machine Learning Scientist in the Trust Detection team, you will design and build innovative machine learning systems to detect and prevent harmful or policy-violating content across the platform. You’ll work on trust, safety, and fraud problems such as phishing prevention, counterfeit detection, and identifying prohibited or restricted listings (e.g. regulated or restricted item categories).

The solutions you build will primarily use large language models and deep learning techniques to operate at scale and with high performance.

• Research, design, and deliver machine learning solutions to detect fraud, abuse, and policy violations in user-generated content
• Work closely with Trust, Product, Policy, and Engineering partners to translate business and safety requirements into effective ML systems
• Build, train, and evaluate LLM-based models for text and multimodal classification, detection, and reasoning tasks
• Set up and run large-scale offline experiments and online evaluations to test hypotheses and measure impact
• Stay up-to-date with research in large language models and modern deep learning, applying new techniques where appropriate
• Participate in team ceremonies including agile rituals, technical design discussions, and roadmap planning
• Clearly communicate technical approaches, results, and trade-offs to both technical and non-technical partners

• Experience working as a Machine Learning Scientist, with a track record of delivering models to solve real-world, production-scale problems
• Strong understanding of machine learning fundamentals, with hands-on experience using frameworks such as PyTorch and modern architectures (e.g. Transformers, large language models)
• Proficiency in Python, with the ability to write production-quality code and a solid understanding of data pipelines, model training, and MLOps practices
• Comfortable working with noisy, weakly-labeled, or imbalanced data typical of trust and safety domains
• Collaborative, pragmatic, and curious teammate, able to work successfully with multi-functional partners
• Passion for learning, experimentation, and keeping up to date with advances in machine learning

• Experience building classification or scoring models for trust, safety, fraud, abuse, or policy enforcement use cases
• Hands-on experience fine-tuning, evaluating, or deploying large language models for real-world applications
• Experience with experiment design, offline evaluation, and online testing (e.g. A/B tests)
• Experience working with Databricks and Py Spark
• Experience deploying ML systems on AWS or other cloud platforms (GCP/Azure)

• PMI and cash plan healthcare access with Bupa
• Subsidised counselling and coaching with Self Space
• Cycle to Work scheme with options from Evans or the Green Commute Initiative
• Employee Assistance Programme (EAP) for 24/7 confidential support
• Mental Health First Aiders across the business for support and signposting

• 25 days annual leave with option to carry over up to 5 days
• 1 company-wide day off per quarter
• Impact hours:
  Up to 2 days additional paid leave per year for volunteering
• Fully paid 4 week sabbatical after completion of 5 years of consecutive service with Depop, to give you a chance to recharge or do something you love.
• Flexible Working:
  MyMode…