Ocelot’s Post

Our Founder, Robert Caulk, is going to present at the Grenoble Data Science Meetup on January 18th, 2024, about how our team is building production-scale retrieval augmented generative applications, using our software flowdapt.ai 😎 . Why did we build flowdapt.ai? To ensure our key priorities were satisfied, including: * 🚲 highly parallelized compute efficiency, * 🤖 automatic resource management, * 🐞 rapid (local) prototyping and debuggability, * 🔌 intuitive cluster-wide data sharing methods, * �� easy scheduling, * 📝 live configurability, and * 🚚 deployment cycle efficiency. It all comes together in our flagship application, asknews.app 🚀 Come learn and grab a beer afterward 🍻 : https://lnkd.in/dZjNMacT

Join us on October 19th for Break into Tech! Learn what an algorithm is, how real-world #data can be represented, and how your point of view can shape computer science or solve a real-world problem using data science. Come explore opportunities and #network with us at our #Miami campus. Register below!

Data leaks are common when hackathons are teaching people to create them. They are also common when AIs are using neurodusts to hack and automate mind and brain data. Hackathons teach people to hack more, better, faster.

📢 I always mention the power of attending to conferences and meet interesting like-minded people. There you go, another great conference for Data lovers with awesome speakers from top tech companies, where you can learn new approaches to modern data problems. Don't forget to register. 📝 See you All there! 😎 #datascience #datafest #dataconference #aicommunity #share

Jay Cui and I have chosen our hackathon project for this Friday. It's going to be a VS Code extension for BentoML. Now I'll pitch the project in hopes we can get a few people to join! ... Even if you're remote! BentoML has facilities for saving ML models in a standard format, and then serving them in all sorts of ways (REST/gRPC API, Spark UDF, Dask, Ray, SageMaker, ...) using an SDK similar to FastAPI. We're making an extension inspired by the Docker extension. (1) BentoML has a CLI that lets you do actions against "models" and "bentos" (a bento is a model + serving code). The UI in the sidebar will provide convenient read functions for those, and let you do actions like "click play to serve this model locally and open the OpenAPI/Swagger docs page". (2) BentoML has some config files, e.g. bentofile .yaml, bento_configuration .yaml, bento_deployment_config .yaml, etc. These let you configure things the dependencies needed by your model, OpenTelemetry traces and metrics, and autoscaling parameters like how many CPUs/GPUs to use in a deployment. We'll register these as a filetype with VS Code and define autocompletion for them! Really excited about this part. Approach: Jordan Pierre and some others created a VS Code extension for ClearML at our last hackathon. This BentoML extension is similar to the ClearML extension in many ways and also simpler. So... we'll copy/paste that extension as a starting place :D This is a great opportunity to contribute to an open-source project that will be appreciated by real users in the ML{Ops|Eng} space. If we can get it to a good place, we'll pitch to BentoML that they adopt it as their official extension. (none currently exists) Skills in any of these areas will be helpful: - TypeScript (or not having a fear of TypeScript/JavaScript as long as GitHub copilot and ChatGPT are at your side) - BentoML - helpful, but it's okay if you haven't used it before I'd recommend studying this repo in advance: https://lnkd.in/gVfFNS2k. It's the VS Code ClearML extension we made. We learned a LOT about creating VS Code extensions in the process of making this one, and we **recorded/demo'd our learnings as we went**. If you watch those recordings, you should come out with loads of useful context. The only "VS Code API" feature that we'll be using in the BentoML extension that we didn't in this reference extension are aspects relating to autocompletion. We'll have a Slack workspace soon. TBD on that. We'll also try to create a repo in advance with the boilerplate code. TBD on that, too. Definitely recommending getting that running on your own machine before the night of. With the ClearML extension, we did most of the work during the hackathon. We had a mixed in-person-remote group. We got really far and had a good time, so we worked async on it here and there for a week afterward and got it ready for the official release. I'm open to that here, too.

We're throwing a SF-themed data hackathon with Modelbit on 2/17 at our SF HQ (⬡ The Hexagon ⬡). What's a data hackathon? It's just a hackathon, but for data! We'll provide unique, curated datasets for you to work on, and teach you how to use Hex and Modelbit for data exploration and machine learning. There'll be presentations and prizes for the best projects. The theme is Infinite City (San Francisco, of course). We’ll have hundreds of interesting SF datasets to dig into— and not just the same old stuff (no poop maps allowed!). Your investigations and insights should be adventures into the unknown, celebrations of the odd and niche, and things that make you look at this marvelous city in a whole new light. space is pretty limited, so please RSVP. See you there 😄 https://lu.ma/7mttq76t

Do you want to be able to build, serve, and deploy machine learning models from your code editor? You might be able to by the end of this week. Even better: you might be one of the folks who make it happen! This Friday (3/8/24), at BYU's annual hackathon, my team is going to build a BentoML VS Code extension that enables machine learning model building, serving, and deployment, all from the VS Code. Want to learn more? Take a look at the slides below and join our team! See you on Friday at the hackathon :) Tagging data folks I've worked with: 🎧 Eric Riddoch Nicolas Bertagnolli Saul Ramirez, Ph.D. Dana Engebretson Chaz Gundry Harry Sullivan Joshua K. Paul McSlarrow James Griffin-Gomez

I am thrilled to share that after four terms, I've completed the Data Science Program at the University of Toronto School of Continuing Studies and the University of Waterloo ! 🎓📊 I only scratched the surface but I am ready to leverage my skills in data analysis, machine learning, and statistics to tackle real-world challenges, especially in the energy sector. ⚡☀️🔋 Let's power the future with data-driven solutions!

🚀 Day 34 of 100 Days Challenge 🚀 Navigating the Graph: Finding If Path Exists 🛤️ The problem at hand challenges us to efficiently identify if there's a path connecting a given source node to a destination node within a graph represented by its edges. 💡 To tackle this challenge, I leveraged the Union-Find data structure, coupled with path compression and union by rank optimizations. Union-Find is a powerful tool for handling disjoint sets and efficiently determining connectivity in a graph. 🤝