Charly La Rocca

PhD in Computer Science

[Charly La Rocca @CharlyRocca]

Hello world

[Charly La Rocca @CharlyRocca]

How do you build the most efficient pipeline network? 🤔 I built a Python app that connects cities using MST & optimizes flow! 🗺️ Visualized with Streamlit/Plotly.

Live App: https://t.co/I8nAlHONCU
Check out the video: https://t.co/uoXEQpT6Xo

[Charly La Rocca @CharlyRocca]

Video: https://t.co/D3SaGWs5OV

[Charly La Rocca @CharlyRocca]

Comparing the SDK from Vercel and Cloudflare to build your custom chatbot.

[Charly La Rocca @CharlyRocca]

Video: https://t.co/BjC2RzZtpa

[Charly La Rocca @CharlyRocca]

Rewrote the MPS parser in C++ to get a 10x speed up

[Charly La Rocca @CharlyRocca]

Video: https://t.co/xvuxX0UykT

[Charly La Rocca @CharlyRocca]

Replacing MPS with Parquet, is it a good idea?

MPS is the standard used to store mathematical problems on disk.
However, MPS is a text-based file format that can be expensive to parse for large instances.
This project benchmarks the runtime required to parse MPS files in MIPLIB using the HiGHS solver.
We propose a new solution to store the problems as parquet files to accelerate the parsing.

[Charly La Rocca @CharlyRocca]

Video: https://t.co/EVqYfOqhF9

[Charly La Rocca @CharlyRocca]

Kafka 4.0 no longer requires Zookeeper. This significantly reduces its overhead and improves its usability.
Here is a demo project to help you get started. We use docker-compose to set up the infrastructure and PySpark to convert the data into Parquet files.

[Charly La Rocca @CharlyRocca]

@fchollet https://t.co/LAXHBdEVFv

[Will @WClemente]

Bitcoin is dead

[Charly La Rocca @CharlyRocca]

@blueleanlogic @JuliaComputing >>> c++

[Charly La Rocca @CharlyRocca]

@TidesofTime RIP Zyzz

[Charly La Rocca @CharlyRocca]

@fchollet Is it just starting?

[Charly La Rocca @CharlyRocca]

@amasad Well it wouldn't be readable either if we used full names. The equations would be too big.

[Charly La Rocca @CharlyRocca]

@Sajjad_Heydari @fchollet Have you tried https://t.co/W2eY0Xf0yH

[Charly La Rocca @CharlyRocca]

@fchollet Might sound impractical. A diagnosis tool based on History object. Example: “99%train_acc/50%val_acc => Warning clear overfitting. We recommend reducing the model size or add regularization, etc.“ Identify bias/variance problems. Guiding the user a bit with common ML knowledge.

[Charly La Rocca @CharlyRocca]

@Docker Thanks! I was waiting for this.

[Vala Afshar @ValaAfshar]

30 years ago (1990), Steve Jobs spoke about using computers to work from home

[Charly La Rocca @CharlyRocca]

@PlayTheBazaar I like the grandiose feeling of #4.