MLOps.community
Demetrios
Relaxed Conversations around getting AI into production, whatever shape that may come in (agentic, traditional ML, LLMs, Vibes, etc)
Episodes
Mentioned books
Oct 21, 2022 • 50min
Voice and Language Tech // Catherin Breslin // Coffee Sessions #129
MLOps Coffee Sessions #129 {Podcast BTS} with Catherin Breslin, Voice and Language Tech co-hosted by Adam Sroka.Join the Community: https://go.mlops.community/YTJoinInGet the newsletter: https://go.mlops.community/YTNewsletter// AbstractBack in the day, Speech Recognition was its own thing. It's a very different flavor of Data Science. You could not use a lot of the tools. It wouldn't cross over to this type of machine learning.Now, with the advancements, Speech Recognition and Machine learning are coming in together. It's interesting to hear right from someone with a Ph.D. level working with some of the biggest companies in the world doing it. The fact that something like Alexa is lots of models back to back and just fathom the complexity of that is quite cool!// BioCatherine is a machine learning scientist and consultant based in Cambridge UK, and the founder of Kingfisher Labs consulting. Since completing her Ph.D. at the University of Cambridge in 2008, Catherine has commercial and academic experience in automatic speech recognition, natural language understanding, and human-computer dialogue systems, having previously worked at Cambridge University, Toshiba Research, Amazon Alexa, and Cobalt Speech. Catherine has been excited by the application of research to real-world problems involving speech and language at scale.// MLOps Jobs board jobs.mlops.community// MLOps Swag/Merchhttps://mlops-community.myshopify.com/// Related Linkswww.catherinebreslin.co.ukhttps://catherinebreslin.medium.com/MLOps Community Newsletter: https://airtable.com/shrx9X19pGTWa7U3YTwitter: https://twitter.com/catherinebuk--------------- ✌️Connect With Us ✌️ ------------- Join our Slack community: https://go.mlops.community/slackFollow us on Twitter: @mlopscommunity Sign up for the next meetup: https://go.mlops.community/registerCatch all episodes, blogs, newsletters, and more: https://mlops.community/Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/Connect with Adam on LinkedIn: https://www.linkedin.com/in/aesrokaConnect with Catherine on LinkedIn: https://www.linkedin.com/in/catherine-breslin-0592423a/Timestamps: [00:00] Catherine's preferred coffee [01:50] Takeaways [03:59] Introduction to Catherine Breslin [05:04] Subscribe to our newsletter! [06:25] Catherine's background [08:13] Speech Recognition trajectory [09:36] Challenges around technologies and tools [11:34] Reflective trend [13:02] Developer experiences hiccups [15:09] Speech Recognition use case backup [16:56] Toshiba research [17:48] Transition from a research lab to working in the industry [20:01] Unit test of Speech Recognition [20:56] Alexa [22:33] Maturity process of Speech Recognition [26:48] Speech Recognition unrecognizing challenges [30:38] Mechanical Terk [33:00] Social media listening [36:48] Development of Speech Recognition, excited about [37:23] Data from people for the Speech Recognition system vs Scowering news vs watching YouTube for a long time [40:00] Disappearing Languages [43:17] Speech-to-speech translation [44:04] Interesting ways to use unfamiliar models to achieve a result [45:40] Meeting transcriptions [48:37] First toy problem of a new Speech Recognition learner [51:37] Kingfisher Labs' problems to tackle [53:38] Translation layer [54:15] Connect with Catherine on Twitter and LinkedIn for available jobs [54:43] Wrap up
Oct 19, 2022 • 45min
Managing Machine Learning Projects // Simon Thompson // MLOps Coffee Sessions #128
MLOps Coffee Sessions #128 with Simon Thompson, Managing Machine Learning Projects, co-hosted by Abi Aryan.// AbstractIt's a cliche to say that choosing and running the algorithms is only a small part of a typical ML project, but despite that, it's true! Setting up and organizing the project, dealing with the data asset, getting to the heart of the business problem, assessing and choosing the models, and integrating them with the business processes in production are all at least as time-consuming and important. Simon has written a book that talks about how these different activities need to be orchestrated and executed, and he hopes that it might be useful for people who are starting out managing ML projects, and help them avoid some of the crunches and catches that seem to trip people up.// BioSimon has been building and running ML projects since 1994 (when he started his Ph.D. in Machine Learning). His first commercial project was for the Royal Navy, and since then, he has worked in Telecom, Defense, Consultancy, Manufacturing, and Finance. This means Simon has experienced a wide range of working environments and different types of projects. As well as working in a variety of commercial environments, Simon collaborated on EU research projects, UK Government-funded research projects, and worked as an industrial rep on three MIT consortia (BigData@CSAIL, Systems That Learn, and the CISR Data Research Board).Simon was also an industrial fellow at the Alan Turing Institute for a year. This means that he has also seen a lot of the communities' practices and concerns as they developed, and he had the chance to put them into use in a commercial environment. Right now, Simon is working for a technology consultancy called GFT, and his job there is primarily to deliver ML projects for companies in the capital markets, such as investment banks, although we also do work in retail banking, insurance, and manufacturing.// MLOps Jobs board jobs.mlops.community// MLOps Swag/Merchhttps://mlops-community.myshopify.com/// Related Linkshttps://medium.com/@sgt101Managing Machine Learning Projects: From design to deployment, by Simon Thompson:https://www.manning.com/books/managing-machine-learning-projectsMLOps Community Newsletter: https://airtable.com/shrx9X19pGTWa7U3YLanguage processing. Simon Thompson CO545 Lecture 10: https://docplayer.net/211236676-Language-processing-simon-thompson-co545-lecture-10.html--------------- ✌️Connect With Us ✌️ -------------Join our Slack community: https://go.mlops.community/slackFollow us on Twitter: @mlopscommunitySign up for the next meetup: https://go.mlops.community/registerCatch all episodes, blogs, newsletters, and more: https://mlops.community/Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/Connect with Abi on LinkedIn: https://www.linkedin.com/in/abiaryan/Connect with Simon on LinkedIn: https://www.linkedin.com/in/simon-thompson-025a7/Timestamps:[00:00] Simon's preferred coffee[00:35] Introduction to co-host Abi Aryan[01:14] Introduction to Simon Thompson[01:21] Takeaways[04:21] Simon's background[04:54] Subscribe to our Newsletter and join our Slack channel![05:40] Managing Machine Learning Projects: From design to deployment book[07:40] Simon's inclination toward computers and engineering[08:59] Simon's first computer project[10:05] Simon's plan for computers and engineering[11:01] Unexpected changes in Machine Learning[12:13] Changes in infrastructure[13:41] Change in open source[16:43] Setting up and organizing Machine Learning Projects[20:57] Requirements before starting Machine Learning Projects[22:57] Management's big challenges[23:53] Fundamental value to get to the money[27:14] Towards ethics[30:50] Ability to scale[32:23] Must-haves and nice-tabs[35:21] Model accuracy and trustworthiness change in different markets[42:45] Free books![44:26] Connect with Simon!
16 snips
Oct 7, 2022 • 1h 2min
Reliable ML // Niall Murphy & Todd Underwood // Coffee Sessions #127
MLOps Coffee Sessions #127 with Niall Murphy & Todd Underwood, Reliable ML co-hosted by David Aponte.Join the Community: https://go.mlops.community/YTJoinInGet the newsletter: https://go.mlops.community/YTNewsletter// AbstractBy applying an SRE mindset to machine learning, authors and engineering professionals Cathy Chen, Kranti Parisa, Niall Richard Murphy, D. Sculley, Todd Underwood, and featured guest authors show you how to run an efficient and reliable ML system. Whether you want to increase revenue, optimize decision-making, solve problems, or understand and influence customer behavior, you'll learn how to perform day-to-day ML tasks while keeping the bigger picture in mind. (Book description from O'Reilly)// BioNiall MurphyNiall has been interested in Internet infrastructure since the mid-1990s. He has worked with all of the major cloud providers from their Dublin, Ireland offices - most recently at Microsoft, where he was the global head of Azure Site Reliability Engineering (SRE). His books have sold approximately a quarter of a million copies worldwide, most notably the award-winning Site Reliability Engineering, and he is probably one of the few people in the world to hold degrees in Computer Science, Mathematics, and Poetry Studies. He lives in Dublin, Ireland, with his wife and two children.Todd UnderwoodTodd is a Director at Google and leads Machine Learning for the Site Reliability Engineering Director. He is also the Site Lead for Google’s Pittsburgh office. ML SRE teams build and scale internal and external ML services and are critical to almost every Product Area at Google. Before working at Google, Todd held a variety of roles at Renesys. He was in charge of operations, security, and peering for Renesys’s Internet intelligence services, which are now part of Oracle's Cloud service. He also did product work for some early social products that Renesys worked on. Before that, Todd was the Chief Technology Officer of Oso Grande, an independent Internet service provider (AS2901) in New Mexico.// MLOps Jobs boardjobs.mlops.community // MLOps Swag/Merchhttps://mlops-community.myshopify.com/// Related LinksReliable Machine Learning book by Cathy Chen, Niall Richard Murphy, Kranti Parisa, D. Sculley, Todd Underwood: https://www.oreilly.com/library/view/reliable-machine-learning/9781098106218/--------------- ✌️Connect With Us ✌️ -------------Join our Slack community: https://go.mlops.community/slackFollow us on Twitter: @mlopscommunitySign up for the next meetup: https://go.mlops.community/registerCatch all episodes, blogs, newsletters, and more: https://mlops.community/Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/Connect with Niall on LinkedIn: https://www.linkedin.com/in/niallm/Connect with Todd on LinkedIn: https ://www.linkedin.com/in/toddunder/Timestamps:[00:00] Niall and Todd's preferred coffee[02:27] Takeaways[06:30] Subscribe to our Video cast and Podcasts channels[11:36] Tips on running different distributed teams[14:40] Burnout prevention[17:14] Process of rewarding people[22:03] Maturity and growth[26:35] Help with tooling[33:44] Magic of different teams' intersections in incident response[40:12] Systemized framing questions[45:47] ML problems[54:21] Preparing the next generation of practitioners[59:27] Official release of the Reliable Machine Learning book[1:01:57] Wrap up
Oct 4, 2022 • 51min
ML Unicorn Start-up Investor Tells-IT-All // George Mathew // MLOps Coffee Sessions #126
MLOps Coffee Sessions #126 with George Mathew, ML Unicorn Start-up Investor, Tells-IT-All.Join the Community: https://go.mlops.community/YTJoinInGet the newsletter: https://go.mlops.community/YTNewsletter// Abstract What's so enticing about enterprise software? It's incredible to see George's idea and vision to invest in generationally enduring companies. Let's look at the way George likes to structure deals with companies while he's reviewing them, and let's look at the MLOps ecosystem through the eyes of the investors.// Bio George Mathew joins Insight Partners as a Managing Director focused on venture stage investments in AI, ML, Analytics, and Data companies as they are establishing product/market Fit. He brings 20+ years of experience developing high-growth technology startups, including most recently being CEO of Kespry. Prior to Kespry, George was President & COO of Alteryx, where he scaled the company through its IPO (AYX). Previously, he held senior leadership positions at SAP and Salesforce.com. He has driven company strategy, led product management and development, and built sales and marketing teams. George holds a Bachelor of Science in Neurobiology from Cornell University and a Master's in Business Administration from Duke University, where he was a Fuqua Scholar.// MLOps Jobs board jobs.mlops.community // MLOps Swag/Merch https://mlops-community.myshopify.com/ // Related Links https://www.insightpartners.com/--------------- ✌️Connect With Us ✌️ ------------- Join our Slack community: https://go.mlops.community/slackFollow us on Twitter: @mlopscommunity Sign up for the next meetup: https://go.mlops.community/registerCatch all episodes, blogs, newsletters, and more: https://mlops.community/ Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/Connect with George on LinkedIn: https://www.linkedin.com/in/gmathew/Timestamps:[00:00] George's preferred coffee[00:13] Introduction and background to George[03:02] Prompting infrastructure[06:43] George deploying startups[11:07] Different areas in the MLOps space[16:18] MLOps is broad[18:46] Factors which George is interested in[23:02] What's fascinating in Enterprise Software[27:17] George's transition story[29:46] Company stage to invest for George[31:51] Evaluating KPIs for deals[39:22] The deals that George had[41:58] George's senses[43:59] Liquidation preferences[45:50] The duration from the first meeting to the term sheet[48:43] Meeting sources[50:03] Wrap up
Sep 30, 2022 • 55min
Databricks Model Serving V2 // Rafael Pierre // Coffee Sessions #125
MLOps Coffee Sessions #125 with Rafael Pierre, Deploying Real-time ML Models in Minutes with Databricks Model Serving V2, co-hosted by Ryan Russon.Join the Community: https://go.mlops.community/YTJoinInGet the newsletter: https://go.mlops.community/YTNewsletter// Abstract From our experience helping customers in the Data and AI field, we learned that the most challenging part of Machine Learning is deploying it. Putting models into production is complex and requires additional pieces of infrastructure as well as specialized people to take care of it - this is especially true if we are talking about real-time REST APIs for serving ML models. With Databricks Model Serving V2, we introduce the idea of Serverless REST endpoints to the platform. This allows teams to easily deploy their ML models in a production-grade platform with a few mouse clicks (or lines of code 😀).// Bio Rafael has worked for 15 years in data-intensive fields within finance in multiple roles: software engineering, product management, data engineering, data science, and machine learning engineering. At Databricks, Rafael has fun bringing all these topics together as a Solutions Architect to help our customers become more and more data-driven.// MLOps Jobs board jobs.mlops.community MLOps Swag/Merch https://mlops-community.myshopify.com/// Related Links https://mlopshowto.comAirflow Summit 2022: https://youtu.be/JsYEOdRBgREING Data Engineering Meetup: https://www.youtube.com/watch?v=gJoxX1rRZJIMLOps World Virtual Summit NYC 2022: https://drive.google.com/file/d/1EXsqmLfrPAsV9i6h6pGfJxVjMO9y6u9a/view?usp=sharing--------------- ✌️Connect With Us ✌️ ------------- Join our Slack community: https://go.mlops.community/slackFollow us on Twitter: @mlopscommunity Sign up for the next meetup: https://go.mlops.community/registerCatch 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/ryanrusson/Connect with Rafael on LinkedIn: https://www.linkedin.com/in/rafaelpierreTimestamps:[00:00] Rafael's preferred coffee[00:14] Intro and background of Ryan[01:18] Takeaways[04:15] Rafael's first job and first touch in data[06:35] IOT with no cloud[08:00] Rafa's personal experience with Kubernetes[09:16] Hardest thing to grasp when starting[10:49] ML in Kubernetes[13:32] Why is Kubernetes a good platform for ML?[15:06] Is Kubernetes always the right answer?[19:30] Recap of Ale Solano's background[20:30] Cultural side of MLOps[23:07] Ryan's experience[25:04] Databricks over Kubernetes[27:20] What to gain from Databricks?[29:28] Best way to structure and promote code[32:13] Rationalizing tradeoffs[34:08] Difference from a mature software org[35:26] Best way to deal with artifacts[38:21] RBAC is a must-have[41:10] Favorite tools for modeling[42:19] Wrap up
Sep 27, 2022 • 11min
Monitoring Unstructured Data // Aparna Dhinakaran & Jason Lopatecki // Lightning Sessions #2
Lightning Sessions #2 with Aparna Dhinakaran, Co-Founder and Chief Product Officer, and Jason Lopatecki, CEO and Co-Founder of Arize. Lightning Sessions is sponsored by Arize// Abstract Monitoring embeddings on unstructured data is not an easy feat, let's be honest. Most of us know what it is but don't understand it one hundred percent. Thanks to Aparna and Jason of Arize for breaking down embedding so clearly. At the end of this Lightning talk, we get to see a demo of how Arize deals with unstructured data and how you can use Arize to combat that.// BioAparna DhinakaranAparna is the Co-Founder and Chief Product Officer at Arize AI, a pioneer and early leader in machine learning (ML) observability. A frequent speaker at top conferences and a thought leader in the space, Dhinakaran was recently named to the Forbes 30 Under 30. Before Arize, Dhinakaran was an ML engineer and leader at Uber, Apple, and TubeMogul (acquired by Adobe). During her time at Uber, she built several core ML Infrastructure platforms, including Michaelangelo.Aparna has a BA from Berkeley's Electrical Engineering and Computer Science program, where she published research with Berkeley's AI Research group. She is on a leave of absence from the Computer Vision Ph.D. program at Cornell University.Jason LopateckiJason is the Co-founder and CEO of Arize AI, a machine learning observability company. He is a garage-to-IPO executive with an extensive background in building marketing-leading products and businesses that heavily leverage analytics. Prior to Arize, Jason was co-founder and chief innovation officer at TubeMogul where he scaled the business into a public company and eventual acquisition by Adobe. Jason has hands-on knowledge of big data architectures, programmatic advertising systems, distributed systems, and machine learning and data processing architectures. In his free time, Jason tinkers with personal machine learning projects as a hobby, with a special interest in unsupervised learning and deep neural networks. He holds an electrical engineering and computer science degree from UC Berkeley.// MLOps Jobs board jobs.mlops.community // Related Linkshttps://arize.com/--------------- ✌️Connect With Us ✌️ -------------Join our Slack community: https://go.mlops.community/slackFollow us on Twitter: @mlopscommunitySign up for the next meetup: https://go.mlops.community/registerCatch all episodes, blogs, newsletters, and more: https://mlops.community/Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/Connect with Aparna on LinkedIn: https://www.linkedin.com/in/aparnadhinakaran/Connect with Jason on LinkedIn: https://www.linkedin.com/in/jason-lopatecki-9509941/Timestamps:[00:00] Introduction to the topic[01:13] Troubleshooting unstructured ML models is difficult[01:40] Challenges with monitoring unstructured data[02:10] What data looks like[03:02] Embeddings are the backbone of unstructured models[03:28] ML teams need a common tool[04:06] What are embeddings?[05:08] The real WHY behind AI[06:41] ML observability for unstructured data[07:08] Index and Monitor every Embedding[08:05] Measuring drift of unstructured data[08:54] Interactive visualizations [09:34] Fix underlying data issue[09:44] Data-centric AI workflow[10:08] Demo of the product[12:48] Wrap up
Sep 21, 2022 • 59min
Trustworthy Machine Learning // Kush Varshney // Coffee Sessions #124
MLOps Coffee Sessions #124 with Kush Varshney, Distinguished Research Staff Member and Manager, IBM Research, Trustworthy Machine Learning, co-hosted by Krishnaram Kenthapadi.Join the Community: https://go.mlops.community/YTJoinInGet the newsletter: https://go.mlops.community/YTNewsletter// Abstract Trustworthy ML is a way of thinking and something to be worked on and operationalized throughout the entire machine learning development lifecycle, starting from the problem specification phase, which should include diverse stakeholders.// Bio Kush R. Varshney was born in Syracuse, New York, in 1982. He received a B.S. degree (magna cum laude) in electrical and computer engineering with honors from Cornell University, Ithaca, New York, in 2004. He received the S.M. degree in 2006 and the Ph.D. degree in 2010, both in electrical engineering and computer science at the Massachusetts Institute of Technology (MIT), Cambridge. While at MIT, he was a National Science Foundation Graduate Research Fellow.Dr. Varshney is a distinguished research staff member and manager with IBM Research at the Thomas J. Watson Research Center, Yorktown Heights, NY, where he leads the machine learning group in the Foundations of Trustworthy AI department. He was a visiting scientist at IBM Research - Africa, Nairobi, Kenya in 2019. He is the founding co-director of the IBM Science for Social Good initiative. He applies data science and predictive analytics to human capital management, healthcare, olfaction, computational creativity, public affairs, international development, and algorithmic fairness, which has led to recognitions such as the 2013 Gerstner Award for Client Excellence for contributions to the WellPoint team and the Extraordinary IBM Research Technical Accomplishment for contributions to workforce innovation and enterprise transformation. He conducts academic research on the theory and methods of trustworthy machine learning. His work has been recognized through best paper awards at the Fusion 2009, SOLI 2013, KDD 2014, and SDM 2015 conferences and the 2019 Computing Community Consortium / Schmidt Futures Computer Science for Social Good White Paper Competition. He self-published a book entitled 'Trustworthy Machine Learning in 2022, available at http://www.trustworthymachinelearning.com. He is a senior member of the IEEE.// MLOps Jobs board jobs.mlops.community // MLOps Swag/Merch https://mlops-community.myshopify.com/ // Related Links--------------- ✌️Connect With Us ✌️ ------------- Join our Slack community: https://go.mlops.community/slackFollow us on Twitter: @mlopscommunity Sign up for the next meetup: https://go.mlops.community/registerCatch all episodes, blogs, newsletters, and more: https://mlops.community/Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/Connect with Krishnaram on LinkedIn: https://www.linkedin.com/in/krishnaramkenthapadiConnect with Kush on LinkedIn: https://www.linkedin.com/in/kushvarshney/Timestamps:[00:00] Kush Varshney's introduction[02:16] Takeaways[07:56] Talking about TRUST[12:44] Trustworthy AI[19:32] Adopting Trustworthy Machine Learning[22:19] Weapons of Math Destruction book - biases[30:34] Problem sources[38:08] Self-publishing a Trustworthy Machine Learning book [43:36] Changing the Nature of AI Research by Professor Subbarao Kambhampati[50:08] IBM research support[52:00] Wrap up
Sep 16, 2022 • 52min
RECOMMENDER SYSTEM: Why They Update Models 100 Times a Day // Gleb Abroskin // MLOps Coffee Sessions #123
MLOps Coffee Sessions #123 with Gleb Abroskin, Machine Learning Engineer at Funcorp, RECOMMENDER SYSTEM: Why They Update Models 100 Times a Day, co-hosted by Jake Noble.Join the Community: https://go.mlops.community/YTJoinInGet the newsletter: https://go.mlops.community/YTNewsletter// AbstractFunCorp was a top 10 app store. It was a very popular app that had a ton of downloads and just memes. They need a recommendation system on top of that. Memes are super tricky because they're user-generated and they evolve very quickly. They're going to live and die by the Recommender System in that product.It's incredible to see FunCorp's maturity. Gleb breaks down the feature store they created and the velocity they have to be able to create a whole new pipeline in a new model and put it into production after only a month!// BioGleb makes models go brrrrr. He doesn't know what is expected in this field, to be honest, but Gleb has experience in deploying a lot of different ML models for CV, speech recognition, and RecSys in a variety of languages (C++, Python, Kotlin), serving millions of users worldwide./ MLOps Jobs board jobs.mlops.communityMLOps Swag/Merchhttps://mlops-community.myshopify.com/// Related LinksPutting a two-layered recommendation system into production - https://medium.com/@FunCorp/putting-a-two-layered-recommendation-system-into-production-b8caaf61393d Practical Guide to Create a Two-Layered Recommendation System - https://medium.com/@FunCorp/practical-guide-to-create-a-two-layered-recommendation-system-5486b42f9f63 Ten Mistakes to Avoid When Creating a Recommendation System - https://medium.com/@FunCorp/ten-mistakes-to-avoid-when-creating-a-recommendation-system-8268ed60aeba Applying Domain-Driven Design and Patterns: With Examples in C# and .NET, 1st Edition by Jimmy Nilsson: https://www.amazon.com/Applying-Domain-Driven-Design-Patterns-Examples/dp/0321268202--------------- ✌️Connect With Us ✌️ -------------Join our Slack community: https://go.mlops.community/slackFollow us on Twitter: @mlopscommunitySign up for the next meetup: https://go.mlops.community/registerCatch all episodes, blogs, newsletters, and more: https://mlops.community/Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/Connect with Jake on LinkedIn: https://www.linkedin.com/in/jakednoble/Connect with Gleb on LinkedIn: https://www.linkedin.com/in/gasabr/Timestamps:[00:00] Introduction to Gleb Abroskin[00:50] Takeaways[05:39] Breakdown of FunCorp teams[06:47] FunCorp's team ratio[07:41] FunCorp team provisions[08:48] Feature Store vision[10:16] Matrix factorization[11:51] Fairly modular, fairly thin infrastructure[12:26] Distinct models with the same feature[13:08] FunCorp's definition of Feature Store[15:10] Unified API[15:55] FunCorp's scaling direction[17:01] Level up as needed[17:38] Future of FunCorp's Feature Store[18:37] Monitoring investment in the space[19:43] Latency for business metrics[21:04] Velocity to production[23:10] 30-day retention struggle[24:45] Back-end business stability[27:49] Recommender systems[30:34] Back-end layer headaches[37:41] Continuous training pipelines produce an artifact[39:33] Worst-case scenario[40:38] Realistic estimation of a new model deployment[41:42] Recommender Systems' future velocity [43:07] A/B Testing launch - no launch decision[46:32] Lightning question[47:08] Wrap up
Sep 9, 2022 • 57min
Scaling Similarity Learning at Digits // Hannes Hapke // Coffee Sessions #122
MLOps Coffee Sessions #122 with Hannes Hapke, Machine Learning Engineer at Digits Financial, Inc., Scaling Similarity Learning at Digits, co-hosted by Vishnu Rachakonda.Join the Community: https://go.mlops.community/YTJoinInGet the newsletter: https://go.mlops.community/YTNewsletter// AbstractMachine Learning in a product is a double-edged sword. It can make a product more useful, but it depends on assumed and strictly defined behavior from users. Hannes walks through the entirety of their machine learning pipeline, how they implemented it, what the elements are, what the learning looks like, and what the tooling looks like. Hannes maps out what good data hygiene looks like, not only from the machine learning perspective, but also down to the software engineering, design, and backend engineering, all the way to the data engineering perspectives.// BioHannes was the first ML engineer at Digits, where he built the MLOPs foundation for their ML team. His interest in production machine learning ranges from building ML pipelines to scaling similarity-based ML to process millions of banking transactions daily. Prior to Digits, Hannes implemented ML solutions for a number of applications, incl. retail, health care, or ERP companies.He co-authored two machine learning books:* Building Machine Learning Pipeline (O'Reilly)* NLP in Action (Manning)// MLOps Jobs board jobs.mlops.community// MLOps Swag/Merchhttps://mlops-community.myshopify.com/// Related Links--------------- ✌️Connect With Us ✌️ -------------Join our Slack community: https://go.mlops.community/slackFollow us on Twitter: @mlopscommunitySign up for the next meetup: https://go.mlops.community/registerCatch all episodes, blogs, newsletters, and more: https://mlops.community/Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/Connect with Vishnu on LinkedIn: https://www.linkedin.com/in/vrachakonda/Connect with Hannes on LinkedIn: https://www.linkedin.com/in/hanneshapke/Timestamps: [00:00] Introduction to Hannes Hapke [01:37] Takeaways [02:40] Design supercharges machine learning [05:48] Building Machine Learning Pipeline book [08:09] Updating the edition [09:37] Abstract away [11:52] Approach of crossover [16:04] Training serving skew [20:42] Tools using continuous integration and deployment [25:25] Human in the loop touch point [27:44] Data backfilling update [30:06] Work and Products of Digits [32:26] Digit Boost [35:30] The first machine learning engineer [39:55] Structured data in good shape, good data processing perspective, concept-educated teams [43:33] Digits is hiring! [43:55] Machine Learning struggles [47:10] Design decision [49:49] Data or machine learning literacy [51:30] Data Hygiene [52:49] Rapid-fire questions [54:47] Wrap up
Sep 6, 2022 • 1h 5min
Bringing DevOps Agility to ML// Luis Ceze // Coffee Sessions #121
MLOps Coffee Sessions #121 with Luis Ceze, CEO and Co-founder of OctoML, Bringing DevOps Agility to ML, co-hosted by Mihail Eric. Join the Community: https://go.mlops.community/YTJoinInGet the newsletter: https://go.mlops.community/YTNewsletter// AbstractThere's something about this idea where people see a future where you don't need to think about infrastructure. You should just be able to do what you do, and infrastructure happens. People understand that there is a lot of complexity underneath the hood, and most data scientists or machine learning engineers start deploying things, and shouldn't have to worry about the most efficient way of doing this.// BioLuis Ceze is Co-Founder and CEO of OctoML, which enables businesses to seamlessly deploy ML models to production, making the most out of the hardware. OctoML is backed by Tiger Global, Addition, Amplify Partners, and Madrona Venture Group. Ceze is the Lazowska Professor in the Paul G. Allen School of Computer Science and Engineering at the University of Washington, where he has taught for 15 years.Luis co-directs the Systems and Architectures for Machine Learning lab (sampl.ai), which co-authored Apache TVM, a leading open-source ML stack for performance and portability that is used in widely deployed AI applications. Luis is also co-director of the Molecular Information Systems Lab (misl.bio), which led pioneering research in the intersection of computing and biology for IT applications such as DNA data storage. His research has been featured prominently in the media, including the New York Times, Popular Science, MIT Technology Review, and the Wall Street Journal. Ceze is a Venture Partner at Madrona Venture Group and leads their technical advisory board.// MLOps Jobs boardjobs.mlops.community MLOps Swag/Merchhttps://mlops-community.myshopify.com/// Related LinksLanding page: https://octoml.ai/The Boys in the Boat: Nine Americans and Their Epic Quest for Gold at the 1936 Berlin Olympics by Daniel James Brown:https://www.amazon.com/Boys-Boat-Americans-Berlin-Olympics/dp/0143125478--------------- ✌️Connect With Us ✌️ -------------Join our Slack community: https://go.mlops.community/slackFollow us on Twitter: @mlopscommunitySign up for the next meetup: https://go.mlops.community/registerCatch all episodes, blogs, newsletters, and more: https://mlops.community/Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/Connect with Mihail on LinkedIn: https://www.linkedin.com/in/mihaileric/Connect with Luis on LinkedIn: https://www.linkedin.com/in/luis-ceze-50b2314/Timestamps:[00:00] Introduction to Luis Ceze[06:28] MLOps does not exist[10:41] Semantics argument[16:25] Parallel programming standpoint[18:09] TVM[22:51] Optimizations[24:18] TVM in the ecosystem[27:10] OctoML's further step[30:42] Value chain[33:58] Mature players[35:48] Talking to SREs and Machine Learning Engineers[36:32] Building OctoML[40:20] My Octopus Teacher[42:15] Environmental effects of Sustainable Machine Learning[44:50] Bridging the gap from OctoML to biological mechanisms[50:02] Programmability[57:13] Academia making an impact[59:40] Rapid-fire questions[1:03:39] Wrap up
