Developing Machine Learning Algorithms in LUN Projects
summary
Describes building machine-learning algorithms for a real-estate product (LUN). Starts from simple heuristic modules that classified whether a contact was an agent or an owner, then moved to dozens of ML models to reduce customer complaints. Uses practical ML deployment within the product to improve accuracy and user experience.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 4:25Why LUN uses ML
- 11:55Fake photos and image classification
- 15:26Switching to CNNs
- 20:20Improving ad grouping
- 23:33Watermark removal
- 26:02ML impact on UX
- 28:50ML model deployment
- 32:00Product quirks
- 32:24ML in outdoor ads
- 37:00FLYR AI Research team
- 37:42Future of interior search
- 41:44Open data and open source
- 42:11Q&A
related talks
22mBefore the Model: How Machine Learning Products Start...
59mA developers guide to Machine Learning
39mDeveloping Machine Learning for Impact
56mDehumanizing Agents: Why Explainability is Crucial in the LLM Era
27mA Look Inside Applied Machine Learning
56mMachine Learning in E‑commerce
21mDemystifying Machine Learning for Programmatic UA & Retargeting
43mApplying Machine Learning in the Mobile World: Experiences and Use Cases