Alexander Ocsa @aocsa

Staff Software Engineer at @voltrondata. open-source: @ApacheArrow, @Pytorch, @blazingsql, @rapidsai, @laboratoriala aocsa.dev Arequipa / Joined May 2011

Tweets

2K
Followers

702
Following

3K
Likes

10K

Lingo.dev @lingodotdev

2 days ago

Linux nailed it 😁👏

119 1K 20K 637K 2K

Download Image

This new DeepMind research shows just how broken vector search is. Turns out some docs in your index are theoretically incapable of being retrieved by vector search, given a certain dimension count of the embedding. Plain old BM25 from 1994 outperforms it on recall. 1/4

93 404 4K 462K 5K

Download Image

Shardul @isharduld

2 weeks ago

Good read for understanding modern day system programming language wars especially from a context of a C programmer

2 13 115 8K 112

Download Image

Vivek Galatage @vivekgalatage

2 weeks ago

CMU's course on Advanced Database Systems is one of the best. New to databases? Start with the introduction - see quoted post. 15721.courses.cs.cmu.edu/spring2024/

Vivek Galatage @vivekgalatage

6 months ago

CMU's course on Advanced Database Systems is one of the best. New to databases? Start with the introduction - see quoted post. 15721.courses.cs.cmu.edu/spring2024/ https://t.co/rIiqRSrH8y

12 34 334 54K 367

Download Image

1 97 689 32K 769

Download Image

Madan Khadka @MadanKhd

3 weeks ago

LLM inference engine using C++ and CUDA from scratch without libraries. andrewkchan.dev/posts/yalm.html

3 126 853 40K 980

Download Image

Jacob Austin @jacobaustin132

3 weeks ago

Today we're putting out an update to the JAX TPU book, this time on GPUs. How do GPUs work, especially compared to TPUs? How are they networked? And how does this affect LLM training? 1/n

38 532 3K 376K 5K

Download Image

ℏεsam @Hesamation

3 weeks ago

I think I found a based Substack on low-level GPU programming by accident. He has some extensive articles on CUDA programming, building LLM inference engines, looking inside GPUs and much more. even the name is cool: "From Scratch". bro.

14 157 2K 115K 3K

Download Image

Conrad Irwin @conradirwin

3 weeks ago

Finally figured it out well enough to write it down. zed.dev/blog/why-llms-…

107 347 3K 303K 4K

Sebastian Raschka @rasbt

4 weeks ago

Wrote up a more detailed analysis of gpt-oss, how far we've come since GPT-2, and how it compares to Qwen3: magazine.sebastianraschka.com/p/from-gpt-2-t…

24 156 793 38K 550

Victoria Slocum @victorialslocum

2 months ago

Looking for a cheaper, open source alternative to agentic RAG? Try multi-vector retrieval with Reason-ModernColBERT. Most search systems compare one big summary vector per document. Multi-vector retrieval is different - it keeps separate vectors for each word or phrase, then…

11 90 505 37K 534

Download Image

Mahesh @gutsOfDarkness8

3 months ago

Concurrent Hash Table with Fine Grained Locking : A thread-safe hash table with per-bucket locks for low contention, suited for high-performance systems like databases or caches.

3 19 225 20K 139

Download Image

Gassel @Mlbot4

5 months ago

@aaditsh github.com/ahujasid/blend…

0 2 12 980 13

Brendan Jowett @jowettbrendan

3 months ago

BREAKING: A new open-source AI just made 3D design radically faster. It’s called PartCrafter and it turns a single 2D image into detailed, geometrically distinct 3D meshes. No manual segmentation. No CAD software. Here’s why it matters:

12 21 98 16K 156

Download Video

Aadit Sheth @aaditsh

5 months ago

This guy literally wired Claude into Blender, and now it’s generating wild 3D scenes

11 35 323 26K 235

Download Video

Andrew Ng @AndrewYNg

3 months ago

There’s a new breed of GenAI Application Engineers who can build more-powerful applications faster than was possible before, thanks to generative AI. Individuals who can play this role are highly sought-after by businesses, but the job description is still coming into focus. Let…

139 817 4K 529K 6K

TuringPost @TheTuringPost

3 months ago

.@nvidia never stops surprising Together with @Cornell they presented Eso-LMs (Esoteric Language Models) — a new kind of LM that combines the best parts of autoregressive (AR) and diffusion models. • It’s the first diffusion-based model that supports full KV caching. • At the…

10 67 323 20K 221

Download Image

Shubham Saboo @Saboo_Shubham_

3 months ago

Anthropic just dropped a free course on building AI Apps with MCP. Learn to connect AI Agents to external data sources like GitHub, Google Docs, local files using MCP. 100% free.

21 383 3K 270K 5K

Download Image

👋 Jan @jandotai

3 months ago

Google has quietly open-sourced a full-stack research agent stack, powered by Gemini and LangGraph. It's capable of multi-step web search, reflection, and synthesis. While not confirmed to match Gemini’s production backend, it's strikingly close. github.com/google-gemini/…

12 282 2K 165K 3K

Daniel Lemire @lemire

4 months ago

x.com/i/article/1925…

5 18 125 14K 123

Mohit Mishra @chessMan786

4 months ago

How Your Code Runs: The Journey of a Program Through the CPU This blog post aims to explain the inner workings of the CPU by exploring its architecture, and execution flow, and illustrating these concepts with a practical code example.