Deeptech Labs’ Post

"Last year, I attended one talk at FOSDEM that completely changed my trajectory for this year." These are sentiments we’ve heard from many of our participants. FOSDEM isn’t just another tech conference. It’s complete chaos (in the best possible way), where groundbreaking AI engineering and innovation thrive. What makes FOSDEM stand out? Imagine this: Walking into a room buzzing with the most passionate minds in open-source and AI engineering - people whose GitHub profiles you’ve admired - and now you get to meet them in person. That’s what makes FOSDEM special for so many It’s a place where hands-on work takes the spotlight, whether you’re: → Combining models → Discussing distillation techniques → Showing your latest technical ideas Does this sound like your crowd? You’ll feel right at home if you’re: → Working directly with models → Creating practical, real-world AI solutions → Interested in how AI works behind the scenes Don’t just attend - PRESENT your work and meet the community! Best part? It’s all FREE. No registration is required. Just show up and dive in to deepen your learning and skills. So why wait? Join us at FOSDEM to shape the AI future. INTERESTED? DM “FOSDEM” in the comments below!

Open vs. closed source? Small & specified vs. large, general-purpose models? Where does quantum computing and decentralization fit in, if at all? These are just a few of the topics I had the pleasure of diving into with Jonathan Frankle, Zhou (Jo) Yu, and Vin Sachidananda, PhD at the Two Sigma Ventures AI Infra 2029 event I moderated. Thanks to all the new and familiar faces in the crowd for joining in and engaging with the panel live! Interested in joining future events where we dive deep into the future of various sectors? Fill out the link below!

Fun meetings and tight deadlines! 💻 We just wrapped up another hackathon at lablab.ai, the #AI for Connectivity Hackathon. Even though we missed out on free access to some LLM models, we still managed to build a really cool project and learned a lot along the way. 🧪 Our project focused on creating a RAG-based application using FAISS vector search and chunked embeddings of chemical data. The goal was to provide tailored answers related to drug discovery, helping both patients and doctors. For patients, it acts as a source of accurate information, and for doctors, it works as an intelligent assistant. Check out this quick demo by one of our team members, Saqib Manzar. 🎥 The hackathon wasn’t without challenges, especially when we realized the deadline was a lot closer than we thought. ⏳ Huge thanks to izza Fatima, Muhammad Ali Murtaza, Azhan Khan, and Kpatcha Essossimna BESSEPOU for their amazing teamwork and making the experience so much fun. 🎯 One of the most interesting parts of this project was ensuring the LLM stayed on topic and didn’t produce irrelevant answers. 1️⃣ Run the App at: https://lnkd.in/dvc88rAd 2️⃣ Our Team Page: https://lnkd.in/diKZdKb3 3️⃣ Our GitHub Code: https://lnkd.in/dryUwwdA

Last weekend, together with Erdy Suryadarma, I had a chance to participate in a DOGE hackathon event hosted by lablab.ai. The challenge was to revolutionize government efficiency using xAI's Grok. We come up with an idea to streamline bureaucratic processes by: - providing citizens an intuitive platform to access government services and track the process, with an AI to assist in each step, - and workflow builder to help the government digitize and automate bureaucratic processes. Check out our submission here! we have the repo link, the demo link, and the pitch deck there! https://lnkd.in/gDxWesU3

What a (Thurs)day at SF Tech Week! 🚀 From AI's role in the enterprise to biotech breakthroughs, Thursday brought the heat. 🔥 💪 Three more days to go! Check out these quick tips on how to make the most of your Tech Week experience:

For any #programmers, #bioengineers, and #computationalbiologists - an interesting initiative by Lux Capital in collaboration with Ginkgo Bioworks, Inc. , Recursion , Enveda Biosciences , EvolutionaryScale , LatchBio and others, providing both datasets and tools for #proteindiscovery, #massspec prediction, functional screens and others, with prizes for the top teams using , composed of 5-15 members each. There is still time to participate and join or form a team - details can be found at https://hackathon.bio #hackathon #datascience

Day 1 of our hackathon at Launch Romania HQ: First brave souls to have joined: Ovi Stoica - VOLLM: Working on AI agents to call you and sell you a new mobile phone data plan. Cosmin-Alin Botoroga - DATA SWEEP: Working on a newsletter (more details soon ;)) Myself: Working on a viral system to share the news we've opened applications for ReaktorX. If any other founders want to join this last 2024 hack/work session, DM me, plenty of space in the office. Best part: You can also get involved from the comfort of your arm chair with warm tea in your hand. We've got questions from the founders, leave your responses in the comments below. From Ovi: What's the longest time you've waited on a customer support call? From Cosmin: Where do you usually get your data for research purposes? From myself: What did you get for Christmas? (except anxiety) See you tomorrow on Day 2.

Are you contemplating an "open source" scalable data science adoption? If so, have you evaluated the Kubeflow-Kubernetes ecosystem for your scaled ML/AI/Deep Learning requirements? Recent enhancements to Kubeflow project features have made adoption more efficient, and its acceptance by the CNCF, along with increased community support, provides assurance for adopters. Meanwhile, CERN (Large Hadron Collider) is utilising Kubeflow to explore the mysteries of the universe - We’re glad to have followed the same path, providing a contemporary data science platform that serves our community. More info. on CERN's adoption:

Detecting Deepfakes Summit is a very welcome initiative. Lots of understandable concerns that deepfakes could soon destroy #trust online & leave everyone vulnerable to industrial levels of #fraud. But the brighter news is that a lot of productive time, effort & resources have been invested in recent years by Yoti & others to effectively detect #deepfakes attacks.