Data Scientist, Classification and Scoring - Lokker.com
Location: REMOTE (North America / Europe, UTC-8 to UTC+2)
Employment Type: Full-time
Source: Hacker News
About Lokker
Lokker is a privacy platform. We continuously analyze websites, applications, third-party technologies, and data flows to identify privacy, security, and consent risks in production. We have built a large proprietary dataset from years of observing real-world digital behavior, and we are hiring a data scientist to help us expand and improve the classification, scoring, and risk intelligence built on it.
About the Role
You will develop and improve models that classify digital activity, identify meaningful privacy and security behaviors, and assess the confidence and significance of findings. You will work closely with engineering, product, and privacy experts, and what you build goes directly into the product.
Requirements
- Experience: 2 to 4 years of production experience in data science and machine learning.
- Technical Skills:
- Strong proficiency in Python and SQL.
- Experience with pandas or Polars.
- Proficiency in scikit-learn and modern classification techniques.
- Familiarity with model evaluation and explainability.
- Practical Experience:
- Training and evaluating supervised classification models on real production data.
- Practical LLM experience beyond simple demonstrations.
- Mindset: Curiosity about how websites and internet technologies actually behave.
- Domain Knowledge: No privacy or adtech background is required; we will teach the domain.
- Timezone: The timezone range (UTC-8 to UTC+2) is a hard requirement. We cannot make an exception.
Application Instructions
- Verification: We assume you used AI to help write your application, as do we. However, we cannot verify claims that exist only in your document. Please ensure your claimed skills are visible somewhere public (e.g., LinkedIn with real dates and role details, GitHub, papers, Kaggle, a writeup, or a talk).
- NDA Work: If your best work is under NDA, please state so and describe it in your own words.
- Process: A human reads every application. There is no keyword filter and no automated rejection.