AI Stories

Our guest today is Başak Eskili, Machine Learning Engineer at Booking.com and C-Founder of Marvelous MLOps. In our conversation, we first dive into MLOps, its key components and how Başak got into the field. We then talk about Marvelous MLOps and her new course: "End to end MLOps with Databricks". Başak finally shares more about her current role at Booking with a focus on building feature stores. If you enjoyed the episode, please leave a 5 star review and subscribe to the AI Stories Youtube channel.From beginner to advanced LLM developer course by Towards AI (use the code AISTORIES10 to get a 10% discount): https://academy.towardsai.net/courses/beginner-to-advanced-llm-dev?ref=63e5e3To learn more about Marvelous MLOps: https://www.marvelousmlops.io/End to End MLOps course with Databricks: https://maven.com/marvelousmlops/mlops-with-databricksFollow Başak on LinkedIn: https://www.linkedin.com/in/ba%C5%9Fak-tu%C4%9F%C3%A7e-eskili-61511b58/Follow Neil on LinkedIn: https://www.linkedin.com/in/leiserneil/  ---(00:00) Intro  (02:18) How Başak Got into AI & MLOps  (06:55) Key Components of MLOps  (12:05) Deploying First ML Model  (15:58) Joining Booking.com  (18:11) Best Practices for Building Scalable and Reliable ML Systems  (23:01) Databricks (27:50) Batch vs. Real-Time Predictions  (31:15) Marvelous MLOps  (33:52) Role at Booking.com  (35:45) Feature Stores (45:45) Career Advice  

Our guest today is Max Mergenthaler, Co-Founder and CEO of Nixtla: one of the most popular libraries for time series forecasting. In this conversation, Max first explains how he got into AI and the lessons he learned from building a couple of tech startups. We then dive into Nixtla and forecasting. Max explains how he founded Nixtla and the different libraries available to build stats, ml and deep learning forecasting algorithms. We also tallk about TimeGPT, Nixtla's closed-source foundation model for time series. We finally discuss the future of the field along with mistakes and best practices when working on forecasting projects. If you enjoyed the episode, please leave a 5 star review and subscribe to the AI Stories Youtube channel.To learn more about Nixtla: https://www.nixtla.io/ Open source librairies (StatsForecast, MLForecast, NeuralForecast): https://www.nixtla.io/open-source TimeGPT: https://github.com/Nixtla/nixtlaFollow Max on LinkedIn: https://www.linkedin.com/in/mergenthaler/Follow Neil on LinkedIn: https://www.linkedin.com/in/leiserneil/  ---(00:00) - Intro(02:00) - How Max got into Data & AI(03:44) - Combining Philosophy with Analytics(09:49) - Lessons from building Startups(14:00) - Founding Nixtla(16:23) - Time Series Forecasting(19:25) - StatsForecast, MLForecast, and NeuralForecast(26:16) - TimeGPT & LLMs for Forecasting(34:30) - Why people love Nixtla(42:34) - Future of Forecasting(45:51) - Mistakes & Best Practices in Forecasting(52:12) - Max’s role as CEO (56:09) - Career Advice

Our guest today is Sebastian Raschka, Senior Staff Research Engineer at Lightning AI and bestselling book author.In our conversation, we first talk about Sebastian's role at Lightning AI and what the platform provides. We also dive into two great open source libraries that they've built to train, finetune, deploy and scale LLMs.: pytorch lightning and litgpt. In the second part of our conversation,  we dig into Sebastian's new book: "Build and LLM from Scratch". We discuss the key steps needed to train LLMs, the differences between GPT-2 and more recent models like Llama 3.1, multimodal LLMs and the future of the field. If you enjoyed the episode, please leave a 5 star review and subscribe to the AI Stories Youtube channel.Build a Large Language Model From Scratch Book: https://www.amazon.com/Build-Large-Language-Model-Scratch/dp/1633437167Blog post on Multimodal LLMs: https://magazine.sebastianraschka.com/p/understanding-multimodal-llmsLightning AI (with pytorch lightning and litgpt repos): https://github.com/Lightning-AIFollow Sebastian on LinkedIn: https://www.linkedin.com/in/sebastianraschka/Follow Neil on LinkedIn: https://www.linkedin.com/in/leiserneil/  ---(00:00) - Intro(02:27) - How Sebastian got into Data & AI(06:44) - Regressions and loss functions(13:32) - Academia to joining LightningAI(21:14) - Lightning AI VS other cloud providers(26:14) - Building PyTorch Lightning & LitGPT(30:48) - Sebastian’s role as Staff Research Engineer(34:35) - Build an LLM From Scratch(45:00) - From GPT2 to Llama 3.1(48:34) - Long Context VS RAG(56:15) - Multimodal LLMs(01:03:27) - Career Advice

Our guest today is Loubna Ben Allal, Machine Learning Engineer at Hugging Face 🤗 . In our conversation, Loubna first explains how she built two impressive code generation models: StarCoder and StarCoder2. We dig into the importance of data when training large models and what can be done on the data side to improve LLMs performance. We then dive into synthetic data generation and discuss the pros and cons. Loubna explains how she built Cosmopedia, a dataset fully synthetic generated using Mixtral 8x7B.Loubna also shares career mistakes, advice and her take on the future of developers and code generation.  If you enjoyed the episode, please leave a 5 star review and subscribe to the AI Stories Youtube channel.Cosmopedia Dataset: https://huggingface.co/blog/cosmopediaStarCoder blog post: https://huggingface.co/blog/starcoderFollow Loubna on LinkedIn: https://www.linkedin.com/in/loubna-ben-allal-238690152/Follow Neil on LinkedIn: https://www.linkedin.com/in/leiserneil/  ---(00:00) - Intro(02:00) - How Loubna Got Into Data & AI(03:57) - Internship at Hugging Face(06:21) - Building A Code Generation Model: StarCoder(12:14) - Data Filtering Techniques for LLMs(18:44) - Training StarCoder(21:35) - Will GenAI Replace Developers? (25:44) - Synthetic Data Generation & Building Cosmopedia(35:44) - Evaluating a 1B Params Model Trained on Synthetic Data(43:43) - Challenges faced & Career Advice