I had a chat with a staff DS at Meta in the Gen AI product and growth team.

It inspired me to pursue product DS, because I believe it's a good skill set to have if I'm building a startup, having product sense is useful.

DS at Meta

• Mission: driving impact through data-informed decision making.
• Misconception: It's not just about using fancy machine learning techniques, but about leveraging data to improve products and user experiences.
• Specialization vs. Full-Stack: While smaller companies might require full-stack data scientists, Meta's scale allows for more specialized roles. This enables deeper expertise in specific areas of data science.

key areas of focus

1. product
   • Use both descriptive and predictive analytics to uncover insights
   • Examples:
     • Analyzing user engagement with Meta AI
     • Determining which countries or languages to prioritize for expansion
     • Deciding on UI elements like whether to show a search bar in WhatsApp
2. ai eval
   • Goal: Create the best AI assistant possible
   • Key metrics: User acquisition, engagement (daily/monthly active users), retention
   • Evaluation methods:
     • Instruction following accuracy
     • Academic benchmarks (continuously evolving)
     • Custom evaluation metrics tailored to Meta's specific needs
3. product-market fit analysis
   • View product adoption as a funnel
   • Key metrics:
     • Adoption rate (e.g., percentage of WhatsApp users using Meta AI)
     • Retention (monthly return rate)
   • Use techniques like A/B testing and holdout tests to measure impact
4. gen ai advancements
   • Develop retrieval mechanisms for real-time knowledge supplementation

essential skills

1. hard skills
   • Strong foundation in statistics and probability
   • Engineering mindset for building and implementing solutions
2. soft skills
   • Effective communication to convey insights and build trust
   • Product sense (especially for product scientists)
   • Analytical thinking and attention to detail
3. passion
   • Think deeply about products and user experiences, obsess over the details
   • Question design decisions and try to reverse-engineer product changes

some resources for Meta DS interviews

• Meta Product Data Scientist
• Meta Data Science Interview Guide [30 LEAKED Questions from 2024]
• Meta Data Science, Analytics Screening: Interview Cheat Sheet | by Priscilla Mannuel
• The Product Data Science Interview Guide