Long, long time.

Latest integration to PyG is the embeddable graph DBMS Kuzu - see how you use Kuzu to build PyG pipelines to leverage complete DBMS capabilities: github.com/kuzudb/kuzu kuzudb.com/blog/kuzu-0.0.…

Why Graph DBMSs Need New Join Algorithms: Story of Worst-case Optimal Joins: lnkd.in/gjpR_8Z2 I wrote a post on "worst-case optimal" join algorithms: a core algorithm in Kùzu & a topic that has kept me busy for long. Tldr; 👇 #kuzudb #graphdatabase #graphdb #gdbms 1/5

Next week we have Semih Salihoğlu (Semih Salihoğlu) talking about "Kùzu Graph Database Management System" Listen from Monday at 8AM GMT. Available on Spotify, Apple and more! Kùzu ➡️ kuzudb.com Support the podcast ☕️⬇️ buymeacoffee.com/disseminate

Thanks unsafe { Jack Waudby }, for hosting me on your podcast! We talked about Kùzu, graph dbms's and some good phd topics to take on in the area! Listen further if you are interested!

Listen to this week's episode with Semih Salihoğlu (Semih Salihoğlu) to learn about Kùzu! Kùzu is an in-process property graph database management system Learn more ➡️ kuzudb.com Support ☕️👇🏼 buymeacoffee.com/disseminate

Looking forward to talking to developers about using Kùzu for graph machine learning applications tomorrow at Microsofts' Toronto office tomorrow!

Still amazed by how students from University of Waterloo - a comparatively small university - place top in the International Collegiate Programming Contest (ICPC) every year Waterloo's Cheriton School of Computer Science Waterloo Mathematics

Who uses DuckDB for real? Very interesting discussion. Seems that DuckDB is gaining widespread popularity in the data science domain. Can we simply use SQL (instead of Python, like Pandas) to do data science??? reddit.com/r/dataengineer…

Sunday rant. For software engineering, my sense is that the phrase “premature optimization is the root of all evil” has massively backfired. Its from a book on data structures and mainly tried to dissuade people from prematurely write things in assembler. But the point was to

Excited to open-source LargeFlock: github.com/dsg-polymtl/la…. An approach to seamlessly mix analytics and semantic analysis using LMs. We introduce on top of TABLE(s) in SQL's DDL PROMPT(s) and MODEL(s). We further add to it DML, functions: "lf_map" + specialized ones (1/n)

FlockMTL is a DuckDB community extension we recently open-sourced (github.com/dsg-polymtl/du…). It enables the seamless combination of analytics and semantic analysis using functions such as llm_complete, llm_complete_json, llm_filter, and llm_embedding (examples as imgs). It

we're turning @uwaterloo into a supercomputer! arceus is a cross-device distributed compute network for training large models, using model/tensor/pipeline parallelism. you can train anything, from deep neural networks to language models on the network. oss & deployment soon!

Thoughts about o3: I'll skip the obvious part (extraordinary reasoning, FrontierMath is insanely hard, etc). I think the essence of o3 is about *relaxing a single-point RL super intelligence* to cover more points in the space of useful problems. The world of AI is no stranger to