barisozmen.github.io

Machine Learning Best Practices

Lessons learned articles

• Amershi et al. 2019. Microsoft. Software Eng for Machine Learning: A Case Study (Microsoft paper)
• Putting ML into production systems (article)
• Data Validation for Machine Learning (Google Research paper)
• Building ML pipeline at AppNexus (blog)
  1. Research-Production Parity
  2. Preventing Pipeline Jungles
  3. Experiments Versioning and Collaboration
  4. Model Freshness Trade-offs
• Applied Machine Learning at Facebook: A Datacenter Infrastructure Perspective (Facebook paper) (a commenting blog)
• Applying Deep Learning to AirBnb Search (KDD’19 paper)

Scaling

• Scaling ML at Uber (Uber Engineering blog)
• Scaling ML as a service (Uber paper)
• Scaling ML productivity at LinkedIn (LinkedIn Engineering blog)

On technical debt

• 2014 NIPS. Sculley et al. Machine Learning: The High Interest Credit Card of Technical Debt (Google AI paper)
• 2015 NIPS. Sculley et al. Hidden Technical Debt in Machine Learning Systems (paper)

Resource Resources

The morning paper (nice blog explaining one ML paper at a day)