Interesting things I’ve read on the internet

• How and why Netflix built a real time distributed graph
• Running comments over email
• AI Systems Engineering Patterns
• Notes on Reinforcement Learning
• Bottom up programming as the root of LLM dev skepticism
• Direct Preference Optimization Explained
• The smol training playbook
• Reinforcement Learning from Human Feedback
• Essential Guide to Hiring a VP/CTO for Startups
• Reflections and Rantings from a System Design Interviewer
• Nvidia’s $20B Antitrust Loophole
• Compressed Filesystems á la Language Models
• Some Reasons to work on Productivity and Velocity
• Quantization Decisions made in the development of Kimi K2 Thinking
• Sam Altman’s Start up Playbook
• Performance Hints
• Cursor Migrating it’s CMS to code
• The continual learning problem
• An efficient way to compute if two triangles overlap
• Activation Aware Weight Quantization
• OpenEvolve Load Balancing Algorithm
• Recursive Language Models
• Your data model is your destiny
• Beating L1 cache with value speculation
• Why do we compare everything to the first piece of information we receive?
• Faster Index Builds and Scalable Queries with GPU CAGRA in Milvus
• The Quick and Dirty Genius of Luhn Algorithm
• Anchored Diffusion Language Model (ADLM)
• MeshSplatting: 3D Gaussian Splatting with Mesh