Tech at Golem.ai

If you need to convert some MSG to EML, create an EML with loads of attachments, or create multiple EML with a single attachment each: try them out here golem.ai/en/file-conver… - nath

We treat a lot of emails, our team regularly needs to create or convert some eml and msg files. @Kaonashi_42 built some tools to help them and we recently made them available for free on our website!

I'm on board with this too! Find ways to reduce time to deploy, the faster it's live, the better! Earlier feedbacks, more useful features, less integrations costs etc etc... That's how we work with the team @TechAtGolemAI 🚀

Identifying straight lines to detect tables in files, allowing us to then read the text inside the table differently 🔎 - @emmanuel_micard , @Nakatoxx

The next two tracks were tests to see how we could improve our text extraction system. 📑 Relying on whitespaces, with clustering and coordinates to group text together instead of using regular line by line text. - Anne-Sophie, Nino, Kevin

🚧Workshop Friday for the tech team at @golem_ai ! 4 different experimentations to see what could be done in one day, a quick tour of what was tried: (open the thread 🧵⬇️) - The Tech Team

"small" query improvement on one of our bigger API calls. We grouped some calls that were not parallelized and made some tweaks to the query so the ORM would fetch all the needed data in a single call instead of relying on lazy loading. Query time down 50% with low efforts!

How it started How it's going

Thanks to a member our team I discovered the keyword infinity, linux only, that you can use with the command sleep. Very useful when trying to get into a kubernetes pod which won’t stop crashing when you use its real command but you want to execute bash inside. - AnneSo

Just discovered that @Scaleway has a Data Center Bunker, I find it oh so cool. Here's their article on the subject: scaleway.com/en/blog/meet-f… - AnneSo

Every month or so we hold a Tech Retro during which we talk about how our work has been doing, tasks we liked or drove us nuts. To help express our feelings and make it entertaining we have an Excalidraw page on which we draw a timeline and add memes whenever we feel like it

@AdrienBrault @DeAlmeidaBenja2 Hello Adrien! The aim of this article was to demonstrate our testing of RAG. A more detailed article is likely to be published in the future.

@TechAtGolemAI @DeAlmeidaBenja2 Hey, is part of the article missing? I see the article basically end with 2 examples without/with RAG. The article seems light on technical details/insights. "The main challenge will be related to" would be nice to learn more about your approach to mentioned challenges.

Hey ! We tried to use the RAG implementation for our product and summarized it here → dev.to/tech_at_golema… If you want to learn about this technology you can read this nice article → scalastic.io/en/mixtral-oll… - @DeAlmeidaBenja2 and Kevin Baude #ai #tech #engineering

Did we just hit 100 deploys this year ??? 🎉🎉