(1/2) After over 7 years, I resigned from Twitter last Friday. I got to work on many things here, but building the Home Timeline was the most fun I have had in my career. A large part of it was the amazing people I had the good fortune to collaborate with and who made me better.
Incredible case study that's readable yet filled with practical approaches and tactics:
✔ Use GEPA to get near-frontier performance at 1/100th the cost.
✔ Quickly evaluate new models, without having to rewrite prompts.
✔ Evolve & maintain your programs w/o full rewrites
💡 Two arXiv papers published in recent days (one from us, one from TUD) reach the same conclusion: LLMs can now generate C++ code for SQL processing that outperforms classical database systems.
⚙️ Our code generator is based on Claude Code and exploits multiple agents working in parallel. Each agent performs tasks typically associated with different components in a #DBMS, such as workload analysis, query optimization, or physical design tuning.
📊 We compare to various classical #DBMS such as DuckDB, ClickHouse, Umbra, MonetDB, and PostgreSQL, finding that the agent-generated code is often significantly faster. Code generation costs are moderate (<$20), making the approach practical for frequently executed queries.
🤖 Analyzing generated code, we find that agents exploit various optimization techniques, including query-specific data structures, as well as low-level optimizations that are specific to the hardware cache hierarchy of our server.
📃 Paper: https://t.co/PBLC9XX5Tv
💾 Code: https://t.co/MvdH8lHT6F
🌐 Site: https://t.co/2WM2gedwbi
@lojil192574 #LLM #Databases #AI #DB
We just rolled out our Spring ’25 release with some big updates to Dropbox Dash, our AI search and knowledge management product. Here’s what’s new:
📷 Search across video, audio, and images — no more digging for cryptic file names. You can even use Dash to search for people you work with.
(1/2)
We just rolled out our Spring ’25 release with some big updates to Dropbox Dash, our AI search and knowledge management product. Here’s what’s new:
📷 Search across video, audio, and images — no more digging for cryptic file names. You can even use Dash to search for people you work with.
(1/2)
(2/2) I leave Twitter with many fond memories and wish the best to the team who will continue taking Twitter to new heights in the challenging times ahead. #lovewhereyouworked 💙
(1/2) After over 7 years, I resigned from Twitter last Friday. I got to work on many things here, but building the Home Timeline was the most fun I have had in my career. A large part of it was the amazing people I had the good fortune to collaborate with and who made me better.
We describe a method for maximising long term engagements by using model-based reinforcement learning. The policy is then used to make decisions about whether to send push notifications or not. More details are in the paper if you are interested. https://t.co/Ww6lKHoim9