Handling Multi-Terabyte LLM Checkpoints // Simon Karasik // #228

Join us at our first in-person conference on June 25 all about AI Quality: https://www.aiqualityconference.com/ Syed Asad is an Innovator, Generative AI & Machine Learning Engineer, and a Champion for Ethical AI MLOps podcast #233 with Syed Asad, Lead AI/ML Engineer at KiwiTech // Retrieval Augmented Generation. A big thank you to @ for sponsoring this episode! AWS - // Abstract Everything and anything around RAG. // Bio Currently Exploring New Horizons: Syed is diving deep into the exciting world of Semantic Vector Searches and Vector Databases. These innovative technologies are reshaping how we interact with and interpret vast data landscapes, opening new avenues for discovery and innovation. Specializing in Retrieval Augmented Generation (RAG): Syed's current focus also includes mastering Retrieval Augmented Generation Techniques (RAGs). This cutting-edge approach combines the power of information retrieval with generative models, setting new benchmarks in AI's capability and application. // MLOps Jobs board https://mlops.pallet.xyz/jobs // MLOps Swag/Merch https://mlops-community.myshopify.com/ // Related Links Website: https://sanketgupta.substack.com/ Our paper on this topic "Generalized User Representations for Transfer Learning": https://arxiv.org/abs/2403.00584 Sanket's blogs on Medium in the past: https://medium.com/@sanket107 --------------- ✌️Connect With Us ✌️ ------------- Join our slack community: https://go.mlops.community/slack Follow us on Twitter: @mlopscommunity Sign up for the next meetup: https://go.mlops.community/register Catch all episodes, blogs, newsletters, and more: https://mlops.community/ Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/ Connect with Syed on LinkedIn: https://www.linkedin.com/in/syed-asad-76815246/

Sanket works as a Senior Machine Learning Engineer at Spotify working on building end-to-end audio recommender systems. Models built by his team are used across Spotify in many different products including Discover Weekly and Autoplay. Join us at our first in-person conference on June 25 all about AI Quality: https://www.aiqualityconference.com/ MLOps podcast #232 with Sanket Gupta, Senior Machine Learning Engineer at Spotify // RecSys at Spotify. A big thank you to LatticeFlow for sponsoring this episode! LatticeFlow - https://latticeflow.ai/ // Abstract LLMs with foundational embeddings have changed the way we approach AI today. Instead of re-training models from scratch end-to-end, we instead rely on fine-tuning existing foundation models to perform transfer learning. Is there a similar approach we can take with recommender systems? In this episode, we can talk about: a) how Spotify builds and maintains large-scale recommender systems, b) how foundational user and item embeddings can enable transfer learning across multiple products, c) how we evaluate this system d) MLOps challenges with these systems // Bio Sanket works as a Senior Machine Learning Engineer on a team at Spotify building production-grade recommender systems. Models built by my team are being used in Autoplay, Daily Mix, Discover Weekly, etc. Currently, my passion is how to build systems to understand user taste - how do we balance long-term and short-term understanding of users to enable a great personalized experience. // MLOps Jobs board https://mlops.pallet.xyz/jobs // MLOps Swag/Merch https://mlops-community.myshopify.com/ // Related Links Website: https://sanketgupta.substack.com/ Our paper on this topic "Generalized User Representations for Transfer Learning": https://arxiv.org/abs/2403.00584 Sanket's blogs on Medium in the past: https://medium.com/@sanket107 --------------- ✌️Connect With Us ✌️ ------------- Join our slack community: https://go.mlops.community/slack Follow us on Twitter: @mlopscommunity Sign up for the next meetup: https://go.mlops.community/register Catch all episodes, blogs, newsletters, and more: https://mlops.community/ Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/ Connect with Sanket on LinkedIn: www.linkedin.com/in/sanketgupta107

Join us at our first in-person conference on June 25 all about AI Quality: https://www.aiqualityconference.com/ Ryan Carson. CEO Founder for 20 years Built and sold 3 startups Helping build a global community of AI devs with Intel. MLOps podcast #231 with Ryan Carson, Senior AI Dev Community Lead at Intel Huge thank you to Zilliz for sponsoring this episode. Zilliz - https://zilliz.com/ // Abstract Ryan shares his professional journey, tracing his transition from building Treehouse to joining Intel. The conversation evolves into a deep dive into Carson's aspiration to democratize access to AI development. Furthermore, he expounds on the exciting prospects of new technology like Gaudi three, a new ASIC for AI workloads. Ryan emphasizes the need for driving competition in compute to lower prices and increase access, underlining the importance of associating individual work with company-based OKRs or KPIs. There is also a reflection on the essentiality of forging quality relationships in professional settings and aligning work with top-level OKRs. Discussion on the potential benefits of AI in constructing and maintaining professional interactions is explored. Touching upon practical applications of AI, they also delve into smaller projects, the possibility of one-person companies, and the role of AI for daily interactions. The episode concludes with an expression of optimism about technological advances shaping the future and an appreciation for the enlightening conversation. // Bio Ryan has been a founder, entrepreneur, and CEO for 20 years, successfully building, scaling, and selling three companies. He's passionate about empowering people to become developers and then connecting them together in a global community. After earning a degree in Computer Science in Colorado, Ryan moved to the UK and worked as a web developer. He then organized global tech conferences, hosting thousands of attendees and influential speakers such as Mark Zuckerberg, the founders of Android, Instagram, and Twitter, among others. His company also produced Twitter’s and Stack Overflow’s developer conferences. Following that, Ryan started an online Computer Science school. Under his leadership, the team grew to over 100 employees, educating more than 1,000,000 students. During this period, he secured $23 million in venture capital and earned recognition as Entrepreneur of the Year. Over the last two years Ryan dove deep into AI and LLMs. He built an educational proof-of-concept called maple.coach, which focuses on teaching Sales. The platform is built using technologies like Next.js, TypeScript, gpt-4, and Vercel. Outside of work, Ryan shares his life with his wife of 20 years and their two teenagers in Connecticut. They enjoy spending their free time sailing and taking walks with their Sheltie, Brinkley. // MLOps Jobs board https://mlops.pallet.xyz/jobs // MLOps Swag/Merch https://mlops-community.myshopify.com/ // Related Links Website: ryancarson.com --------------- ✌️Connect With Us ✌️ ------------- Join our slack community: https://go.mlops.community/slack Follow us on Twitter: @mlopscommunity Sign up for the next meetup: https://go.mlops.community/register Catch all episodes, blogs, newsletters, and more: https://mlops.community/ Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/ Connect with Ryan on LinkedIn: https://www.linkedin.com/in/ryancarson/

Salman Avestimehr is a Dean's Professor, the inaugural director of the USC-Amazon Center for Secure and Trusted Machine Learning (Trusted AI), and director of the Information Theory and Machine Learning (vITAL) research lab. He is also the CEO and co-founder of FedML. MLOps podcast #230 with Salman Avestimehr, CEO & Founder of FedML, FedML Nexus AI: Your Generative AI Platform at Scale. A big thank you to FEDML for sponsoring this episode! // Abstract FedML is your generative AI platform at scale to enable developers and enterprises to build and commercialize their own generative AI applications easily, scalably, and economically. Its flagship product, FedML Nexus AI, provides unique features in enterprise AI platforms, model deployment, model serving, AI agent APIs, launching training/Inference jobs on serverless/decentralized GPU cloud, experimental tracking for distributed training, federated learning, security, and privacy. // Bio Salman is a professor, the inaugural director of the USC-Amazon Center for Secure and Trusted Machine Learning (Trusted AI), and the director of the Information Theory and Machine Learning (vITAL) research lab at the Electrical and Computer Engineering Department and Computer Science Department of the University of Southern California. Salman is also the co-founder and CEO of FedML. He received his Ph.D. in Electrical Engineering and Computer Sciences from UC Berkeley in 2008. Salman does research in the areas of information theory, decentralized and federated machine learning, secure and privacy-preserving learning, and computing. // MLOps Jobs board https://mlops.pallet.xyz/jobs // MLOps Swag/Merch https://mlops-community.myshopify.com/ // Related Links https://www.avestimehr.com/ https://fedml.ai/ --------------- ✌️Connect With Us ✌️ ------------- Join our slack community: https://go.mlops.community/slack Follow us on Twitter: @mlopscommunity Sign up for the next meetup: https://go.mlops.community/register Catch all episodes, blogs, newsletters, and more: https://mlops.community/ Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/ Connect with Salman on LinkedIn: https://www.linkedin.com/company/fedml/ Timestamps: [00:00] AI Quality: First in-person conference on June 25 [01:28] Salman's preferred coffee [01:49] Takeaways [03:33] Please like, share, leave a review, and subscribe to our MLOps channels! [03:53] Challenges that inspired Salman's work [06:20] Controlled ownership [08:11] Dealing with data leakage and privacy problems [10:45] In-house ML Model Deployment [13:36] FEDML: Comprehensive Model Deployment [17:27] Integrating FEDML with Kubernetes [19:46] AI Evaluation Trends [24:37] Enhancing NLP with ML [25:48] FEDML: Canary, A/B, Confidence [29:36] FEDML customers [33:21] On-premise platform for secure data management [37:16] Future prediction: data's crucial for better applications [38:18] Maturity in evaluating and improving steps [41:38] Focus on ownership [45:12] Benefits of smaller models for specific use cases [48:57] Verify sensitive tasks, trust quick, important mobile content creation [51:50] Wrap up

Mohamed Elgendy is the Co-Founder & CEO at Kolena. Additionally, Mohamed Elgendy has had 1 past job as the Director Of Product and Engineering at Synapse Technology Corporation. Join us at our first in-person conference on June 25 all about AI Quality: https://www.aiqualityconference.com/ MLOps podcast #228 with Mohamed Elgendy, Co-founder & CEO of Kolena Inc., What is AI Quality? // Abstract Delve into the multifaceted concept of AI Quality. Demetrios and Mo explore the idea that AI quality is dependent on the specific domain, equitable to the difference in desired qualities between a $1 pen and a $100 pen. Mo underscores the performance of a product being in sync with its intended functionality and the absence of unknown risks as the pillars of AI Quality. They emphasize the need for comprehensive quality checks and adaptability of standards to differing product traits. Issues affecting edge deployments like latency are also highlighted. A deep dive into the formation of gold standards for AI, the nuanced necessities for various use cases, and the paramount need for collaboration among AI builders, regulators, and infrastructure firms form the core of the discussion. Elgendy brings to light their ambitious AI Quality Conference, aiming to set tangible, effective, but innovation-friendly Quality standards for AI. The dialogue also accentuates the urgent need for diversification and representation in the tech industry, the variability of standards and regulations, and the pivotal role of testing in AI and machine learning. The episode concludes with an articulate portrayal of how enhanced testing can streamline the entire process of machine learning. // Bio Mohamed is the Co-founder & CEO of Kolena and the author of the book “Deep Learning for Vision Systems”. Previously, he built and managed AI/ML organizations at Amazon, Twilio, Rakuten, and Synapse. Mohamed regularly speaks at AI conferences like Amazon's DevCon, O'Reilly's AI conference, and Google's I/O. // MLOps Jobs board https://mlops.pallet.xyz/jobs // MLOps Swag/Merch https://mlops-community.myshopify.com/ // Related Links Website: www.kolena.io Deep Learning for Vision Systems book: https://www.amazon.com/Learning-Vision-Systems-Mohamed-Elgendy/dp/1617296198/ --------------- ✌️Connect With Us ✌️ ------------- Join our slack community: https://go.mlops.community/slack Follow us on Twitter: @mlopscommunity Sign up for the next meetup: https://go.mlops.community/register Catch all episodes, blogs, newsletters, and more: https://mlops.community/ Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/ Connect with Mo on LinkedIn: https://www.linkedin.com/in/moelgendy/ Timestamps: [00:00] Mo's preferred coffee [00:07] Takeaways [02:52] See you all in San Francisco on June 25! [03:04] Please like, share, leave a review, and subscribe to our MLOps channels! [03:22] AI Quality in Mo's eyes [08:36] Quality Standards for Software [14:11] Common Chatbot Functionality [19:20] The Birth of Innovation [24:27] Transforming Insights into Standards [30:27] Testing: One step to quality [34:58] Two different data points to be harmonized [37:29] Model cards [39:12] Test Coverage Democratizes Collaboration [42:55] Representation matters [44:50] Wrap up

Join us at our first in-person conference on June 25 all about AI Quality: https://www.aiqualityconference.com Simon Karasik⁠ is a proactive and curious ML Engineer with 5 years of experience. Developed & deployed ML models at WEB and Big scale for Ads and Tax. Huge thank you to Nebius AI for sponsoring this episode. Nebius AI - https://nebius.ai/ MLOps podcast #228 with Simon Karasik, Machine Learning Engineer at Nebius AI, Handling Multi-Terabyte LLM Checkpoints. // Abstract The talk provides a gentle introduction to the topic of LLM checkpointing: why is it hard, how big are the checkpoints. It covers various tips and tricks for saving and loading multi-terabyte checkpoints, as well as the selection of cloud storage options for checkpointing. // Bio Full-stack Machine Learning Engineer, currently working on infrastructure for LLM training, with previous experience in ML for Ads, Speech, and Tax. // MLOps Jobs board https://mlops.pallet.xyz/jobs // MLOps Swag/Merch https://mlops-community.myshopify.com/ // Related Links --------------- ✌️Connect With Us ✌️ ------------- Join our slack community: https://go.mlops.community/slack Follow us on Twitter: @mlopscommunity Sign up for the next meetup: https://go.mlops.community/register Catch all episodes, blogs, newsletters, and more: https://mlops.community/ Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/ Connect with Simon on LinkedIn: https://www.linkedin.com/in/simon-karasik/ Timestamps: [00:00] Simon preferred beverage [01:23] Takeaways [04:22] Simon's tech background [08:42] Zombie models garbage collection [10:52] The road to LLMs [15:09] Trained models Simon worked on [16:26] LLM Checkpoints [20:36] Confidence in AI Training [22:07] Different Checkpoints [25:06] Checkpoint parts [29:05] Slurm vs Kubernetes [30:43] Storage choices lessons [36:02] Paramount components for setup [37:13] Argo workflows [39:49] Kubernetes node troubleshooting [42:35] Cloud virtual machines have pre-installed mentoring [45:41] Fine-tuning [48:16] Storage, networking, and complexity in network design [50:56] Start simple before advanced; consider model needs. [53:58] Join us at our first in-person conference on June 25 all about AI Quality

Sol Rashidi is an esteemed executive, leader, and influencer within the AI, Data, and Technology space.  Having helped IBM launch Watson in 2011 as one of the earliest world applications of Artificial Intelligence, Sol has pioneered some of the early advancements of space.  Join us at our first in-person conference on June 25 all about AI Quality: https://www.aiqualityconference.com/ Huge thank you to  @WeightsBiases  for sponsoring this episode. WandB Free Courses - http://wandb.me/courses_mlops MLOps podcast #227 with Sol Rashidi, CEO & Co-Founder of ExecutiveAI, Leading Enterprise Data Teams. // Abstract In the dynamic landscape of MLOps and data leadership, Sol shares invaluable insights on building successful teams and driving impactful projects. In this podcast episode, Sol delves into the importance of prioritizing relationships, introduces a pragmatic "Wrong Use Cases Formula" to streamline project prioritization, and emphasizes the critical role of effective communication in data leadership. Her wealth of experience and practical advice provide a roadmap for navigating the complexities of MLOps and leading data-driven initiatives to success. // Bio With eight (8) patents granted, 21 filed, and received awards that include: "Top 100 AI People" 2023 "The Top 75 Innovators of 2023" "Top 65 Most Influential Women in 2023" "Forbes AI Maverick of the 21st Century" 2022 “Top 10 Global Women in AI & Data”, 2023 "Top AI 100 Award", 2023 “50 Most Powerful Women in Tech”, 2022 “Global 100 Power List” - 2021, 2022, 2023 “Top 20 CDOs Globally” - 2022 "Chief Analytics Officer of the Year" - 2022 "Isomer Innovators of the Year" - 2021, 2022, 2023 "Top 100 Innovators in Data & Analytics” - 2020, 2021, 2022, 2023 "Top 100 Women in Business" - 2022 Sol is an energetic business executive and a goal-oriented technologist, skilled at coupling her technical acumen with story-telling abilities to articulate business value with both startups and Fortune 100's who are leaning into data, AI, and technology as a competitive advantage while wanting to preserve the legacy in which they were founded upon. Sol has served as a C-Suite member across several Fortune 100 & Fortune 500 companies including: Chief Analytics Officer - Estee Lauder Chief Data & Analytics Officer - Merck Pharmaceuticals EVP, Chief Data Officer - Sony Music Chief Data & AI Officer - Royal Caribbean Cruise Lines Sr. Partner leading the Digital & Innovation Practice- Ernsty & Young Partner leading Watson Go-To-Market & Commercialization - IBM Sol now serves as the CEO of ExecutiveAI LLC. A company dedicated to democratizing Artificial Intelligence for Humanity and is considered an outstanding and influential business leader who is influencing the space traveling the world as a keynote speaker, and serving as the bridge between established Gen1.0 markets and those evolving into 4.0. // MLOps Jobs board https://mlops.pallet.xyz/jobs // MLOps Swag/Merch https://mlops-community.myshopify.com/ // Related Links Sol's Book will be out on April 30, 2024 Your AI Survival Guide: Scraped Knees, Bruised Elbows, and Lessons Learned from Real-World AI Deployments: https://www.amazon.com/Your-Survival-Guide-Real-World-Deployments/dp/1394272634?ref_=ast_author_mpb --------------- ✌️Connect With Us ✌️ ------------- Join our slack community: https://go.mlops.community/slack Follow us on Twitter: @mlopscommunity Sign up for the next meetup: https://go.mlops.community/register Catch all episodes, blogs, newsletters, and more: https://mlops.community/ Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/ Connect with Sol on LinkedIn: https://www.linkedin.com/in/sol-rashidi-a672291/

Join us at our first in-person conference on June 25 all about AI Quality: https://www.aiqualityconference.com/ Chad Sanderson is passionate about data quality, and fixing the muddy relationship between data producers and consumers. He is a former Head of Data at Convoy, a LinkedIn writer, and a published author. He lives in Seattle, Washington, and is the Chief Operator of the Data Quality Camp. Huge thank you to @amazonwebservices for sponsoring this episode. AWS - https://aws.amazon.com/ MLOps podcast #226 with Chad Sanderson, CEO & Co-Founder of Gable, The Rise of Modern Data Management. // Abstract In this session, Chad Sanderson, CEO of Gable.ai and author of the upcoming O’Reilly book: "Data Contracts," tackles the necessity of modern data management in an age of hyper iteration, experimentation, and AI. He will explore why traditional data management practices fail and how the cloud has fundamentally changed data development. The talk will cover a modern application of data management best practices, including data change detection, data contracts, observability, and CI/CD tests, and outline the roles of data producers and consumers. Attendees will leave with a clear understanding of modern data management's components and how to leverage them for better data handling and decision-making. // Bio Chad Sanderson, CEO of Gable.ai, is a prominent figure in the data tech industry, having held key data positions at leading companies such as Convoy, Microsoft, Sephora, Subway, and Oracle. He is also the author of the upcoming O'Reilly book, "Data Contracts” and writes about the future of data infrastructure, modeling, and contracts in his newsletter “Data Products.” // MLOps Jobs board https://mlops.pallet.xyz/jobs // MLOps Swag/Merch https://mlops-community.myshopify.com/ // Related Links AWS Trainium and Inferentia: https://aws.amazon.com/machine-learning/trainium/ https://aws.amazon.com/machine-learning/inferentia/ --------------- ✌️Connect With Us ✌️ ------------- Join our slack community: https://go.mlops.community/slack Follow us on Twitter: @mlopscommunity Sign up for the next meetup: https://go.mlops.community/register Catch all episodes, blogs, newsletters, and more: https://mlops.community/ Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/ Connect with Chad on LinkedIn: https://www.linkedin.com/in/chad-sanderson/

Patrick Beukema has a Ph.D. in neuroscience and has worked on AI models for brain decoding, which analyzes the brain's activity to decipher what people are seeing and thinking. Join us at our first in-person conference on June 25 all about AI Quality: https://www.aiqualityconference.com/ Huge thank you to LatticeFlow for sponsoring this episode. LatticeFlow - https://latticeflow.ai/ MLOps podcast #225 with Patrick Beukema, Head / Technical Lead of the Environmental AI, Applied Science Organization at AI2, Beyond AGI, Can AI Help Save the Planet? // Abstract AI will play a central role in solving some of our greatest environmental challenges. The technology that we need to solve these problems is in a nascent stage -- we are just getting started. For example, the combination of remote sensing (satellites) and high-performance AI operating at a global scale in real-time unlocks unprecedented avenues to new intelligence. MLOPs is often overlooked on AI teams, and typically there is a lot of friction in integrating software engineering best practices into the ML/AI workflow. However, performance ML/AI depends on extremely tight feedback loops from the user back to the model that enables high iteration velocity and ultimately continual improvement. We are making progress but environmental causes need your help. Join us fight for sustainability and conservation. // Bio Patrick is a machine learning engineer and scientist with a deep passion for leveraging artificial intelligence for social good. He currently leads the environmental AI team at the Allen Institute for Artificial Intelligence (AI2). His professional interests extend to enhancing scientific rigor in academia, where he is a strong advocate for the integration of professional software engineering practices to ensure reliability and reproducibility in academic research. Patrick holds a Ph.D. from the Center for Neuroscience at the University of Pittsburgh and the Center for the Neural Basis of Cognition at Carnegie Mellon University, where his research focused on neural plasticity and accelerated learning. He applied this expertise to develop state-of-the-art deep learning models for brain decoding of patient populations at a startup, later acquired by BlackRock. His earlier academic work spanned research on recurrent neural networks, causal inference, and ecology and biodiversity. // MLOps Jobs board https://mlops.pallet.xyz/jobs // MLOps Swag/Merch https://mlops-community.myshopify.com/ // Related Links Variety of relevant papers/talks/links on Patrick's website: https://pbeukema.github.io/ --------------- ✌️Connect With Us ✌️ ------------- Join our slack community: https://go.mlops.community/slack Follow us on Twitter: @mlopscommunity Sign up for the next meetup: https://go.mlops.community/register Catch all episodes, blogs, newsletters, and more: https://mlops.community/ Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/ Connect with Patrick on LinkedIn: https://www.linkedin.com/in/plbeukema/ Timestamps: [00:00] AI Quality Conference [01:29] Patrick's preferred coffee [02:00] Takeaways [04:14] Learning how to learn journey [07:04] Patrick's day to day [08:39] Environmental AI [11:07] Environmental AI models [14:35] Nature Inspires Scientific Advances [18:11] R&D [24:58] Iterative Feedback-Driven Development [26:37 - 28:07] LatticeFlow Ad [33:58] Balancing Metrics for Success [38:16] Model Retraining Pipeline [44:11] Series Models: Versatility [45:57] Edge Models Enhance Output [50:22] Custom Models for Specific Data [53:53] Wrap up

Join us at our first in-person conference on June 25 all about AI Quality: https://www.aiqualityconference.com/ Verena Weber believes that GenAI is going to transform the way we work and interact with devices. Her mission is to help companies prepare for this transformation. She has strong expertise in NLP and over 7 years of experience in Machine Learning. Huge thank you to  @zilliz  for sponsoring this episode. Zilliz - https://zilliz.com/ MLOps podcast #224 with Verena Weber, Generative AI Consultant at Verena Weber, GenAI in Production - Challenges and Trends. // Abstract The goal of this talk is to provide insights into challenges for Generative AI in production as well as trends aiming to solve some of these challenges. The challenges and trends Verena see are: Model size and moving towards mixture of experts architectures context window - new breakthroughs for context lengths from unimodality to multimodality, next step large action models? regulation in form of the EU AI Act Verena uses the differences between Gemini 1.0 and Gemini 1.5 to exemplify some of these trends. // Bio Verena leverages GenAI in natural language to elevate business competitiveness and navigate its transformative impact. Her varied experience in multiple roles and sectors underpins her ability to extract business value from AI, blending deep technical expertise with strong business acumen. Post-graduation, she consulted in Data Science at Deloitte and then advanced her skills in NLP, Deep Learning, and GenAI as a Research Scientist at Alexa team, Amazon. Passionate about gender diversity in tech, she mentors women to thrive in this field. // MLOps Jobs board https://mlops.pallet.xyz/jobs // MLOps Swag/Merch https://mlops-community.myshopify.com/ // Related Links Website: verenaweber.de Sign up for Verena's newsletter: https://verenas-newsletter-63558b.beehiiv.com/ Zilliz - https://zilliz.com/ --------------- ✌️Connect With Us ✌️ ------------- Join our slack community: https://go.mlops.community/slack Follow us on Twitter: @mlopscommunity Sign up for the next meetup: https://go.mlops.community/register Catch all episodes, blogs, newsletters, and more: https://mlops.community/ Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/ Connect with Verena on LinkedIn: https://www.linkedin.com/in/verena-weber-134178b9/ Timestamps: [00:00] AI Quality Conference [01:33] Verena's preferred coffee [02:15] Takeaways [06:33] Ski Person of Influence [11:31] Verena's background in the last 5-10 years [14:24] Tech Evolution: Rapid Transformation [18:13] Working at Amazon and key challenges [20:10] Research-inspired suggestions [22:21] AI Updates Impact Workflows [22:52] Alexa Query Distribution Analysis [24:06] Innovative Solutions for Alexa [25:27] Robust T5 Data Prompting [27:38] Audio Data Quality Challenges [28:21-29:28] Zilliz ad [29:28] Alexa data transcription and data cleaning [35:38] Considering needs, costs, and complexity [37:44] ChatGPt is not ideal for classification [39:32] Comparison of model building using TF, IDF [45:08] Struggle to boost diversity in conference speakers [47:30] Creating safe environments helps underrepresented individuals participate [48:29] Wrap up