Suggestions
Jacopo Tagliabue
Improving my backhand volley
Jacopo Tagliabue is the co-founder and CTO of Bauplan, a technology company founded in 2022 and based in New York.23 He has a diverse educational background, having studied at various institutions including UNISR, SFI, and MIT.13
Prior to founding Bauplan, Tagliabue was the co-founder and CTO of Tooso, an AI startup in San Francisco that was acquired by Coveo.1 At Coveo, he led the AI team from scale-up to IPO and established Coveo Labs, an applied R&D practice known for its open-source contributions.13
Tagliabue's career spans multiple fields, including Information Retrieval, Data Science, Artificial Intelligence, and Computer Systems.1 He has collaborated with major companies and academic institutions, published research papers, and received awards for his work, including a Best Paper Award at NAACL 21.1
Currently, while building Bauplan, Tagliabue also serves as an Adjunct Professor of Machine Learning at NYU.13 He is actively involved in the academic community, co-organizing conferences and contributing to various research events.1
Tagliabue's LinkedIn profile can be found under the username jacopotagliabue, where he is listed as being based in New York.4
Highlights
Last year, I picked up again model theory after 10-ish years and was surprised by how little I remembered, and how fun it all was.
While unknown to me the first time I studied non-standard analysis, it turns out that the original Leibniz-like calculus is, 300 years later, pretty useful for computers to calculate derivates, a mostly boring topic that today is mostly mentioned under the "doing #AI" umbrella. So what do Leibniz and automatic differentiation have in common?
I decided to write a short and hand-wavy post about it, with some tiny Python code to convince mostly myself that things actually work as the math says: after almost a year, I found the time to edit it for Cantor's Paradise and I'm happy it's now out.
Thanks to my friends - and significantly better math people - Federico Bianchi, Patrick John Chia, Ethan Rosenthal, Chris White, Andrea Polonioli, and Ryan Vilim for helpful comments and WTF-reactions on a previous draft. The final version is certainly “a little bit” better thanks to them.
#deeplearning #logic #math #funfacts
Exactly a month after EvalRS @ #kdd2023, Federico Bianchi explanation of the new #reclist API is as timely as ever - including third-party integrations and some #llm stuff (like the cool kids do 😎 ).
We learned quite a few things running an #opensource project, and organizing the #evalRS workshop was a fantastic way to get users involved: if you do #opensource for #ml and want to exchange notes, please get in touch!
Thanks Towards Data Science for publishing the follow-up to our original article, Federico Bianchi and Patrick John Chia for the outstanding refactoring, Unnati P. for the great work to push the beta out - and of course thanks to our 2022 sponsors, Comet, neptune.ai and Gantry!
Surprise surprise, I'll be giving a (last?) talk on #RecSys evaluation soon enough: stay tuned and check out our #opensource stuff!
Workshop: https://lnkd.in/eZTMn6uk Repo: https://lnkd.in/eXiUgkr7
#recsys #kdd2023 #evaluation
