The Not-so-Secret Ingredient for Successful Machine Learning: Software Engineering
summary
ASAPP’s product core is AI, mostly machine learning, and the talk explains that software engineering is the central component for turning daily ML improvement ideas into production and successful products. It describes challenges that ML-based systems introduce compared to traditional software despite decades of engineering history, and shares day-to-day practices and lessons learned from ASAPP’s Machine Learning Engineering team.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 1:26Model challenges in production
- 2:50ML as cost reduction
- 5:39Production vs training models
- 7:47ML vs traditional software
- 10:34ML system complexities
- 12:39System dependencies in ML
- 15:27Serving models in production
- 17:35Data and training pipelines
- 20:20Training pipeline quality
- 21:44Testing ML systems
- 23:09Experiment strategies
- 26:00Conclusions and takeaways
- 27:47Q&A
related talks
35mCan Data & ML really learn from software engineering?
01h 21mIntroduction to Machine Learning for Software Engineers
13mDesigning AI-Intensive Applications
27mA Look Inside Applied Machine Learning
16mKey Points and Lessons Learned in AI Projects in Mechanical Engineering
35mML & I
41mCan Data & ML really learn from software engineering?
33mAI with a DevOps mindset - experimentation, sharing ..