Listen Now

Alex Watson is the co-founder and CEO of Gretel.ai, a startup that offers APIs for creating anonymized and synthetic datasets. Previously he was the founder of Harvest.ai, whose product Macie, an analytics platform protecting against data breaches, was acquired by AWS. Learn more about Alex and Gretel AI: http://gretel.ai Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: https://www.cyou.ai/newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 02:15 Introducing Alex Watson 03:45 How Alex was first exposed to programming 05:00 Alex's experience starting Harvest AI, getting acquired by AWS, and integrating their product at massive scale 21:20 How Alex first saw the opportunity for Gretel.ai 24:20 The most exciting use-cases for synthetic data 28:55 Theoretical guarantees of anonymized data with differential privacy 36:40 Combining pre-training with synthetic data 38:40 When to anonymize data and when to synthesize it 41:25 How Gretel's synthetic data engine works 44:50 Requirements of a dataset to create a synthetic version 49:25 Augmenting datasets with synthetic examples to address representation bias 52:45 How Alex recommends teams get started with Gretel.ai 59:00 Expected accuracy loss from training models on synthetic data 01:03:15 Biggest surprises from building Gretel.ai 01:05:25 Organizational patterns for protecting sensitive data 01:07:40 Alex's vision for Gretel's data catalog 01:11:15 Rapid fire questions Links: Gretel.ai Blog NetFlix Cancels Recommendation Contest After Privacy Lawsuit Greylock - The Github of Data Improving massively imbalanced datasets in machine learning with synthetic data Deep dive on generating synthetic data for Healthcare Gretel’s New Synthetic Performance Report The...

Radek Osmulski is a fully self-taught machine learning engineer. After getting tired of his corporate job, he taught himself programming and started a new career as a Ruby on Rails developer. He then set out to learn machine learning. Since then, he's been a Fast AI International Fellow, become a Kaggle Master, and is now an AI Data Engineer on the Earth Species Project. Learn more about Radek: https://www.radekosmulski.com https://twitter.com/radekosmulski Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: http://cyou.ai/newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 02:15 How Radek got interested in programming and computer science 09:00 How Radek taught himself machine learning 26:40 The skills Radek learned from Fast AI 39:20 Radek's recommendations for people learning ML now 51:30 Why Radek is writing a book 01:01:20 Radek's work at the Earth Species Project 01:10:15 How the ESP collects animal language data 01:21:05 Rapid fire questions Links: Radek's Book "Meta-Learning" Andrew Ng ML Coursera Fast AI Universal Language Model Fine-tuning for Text Classification How to do Machine Learning Efficiently NPR - Two Heartbeats a Minute Earth Species Project A Guide to the Good Life The Origin of Wealth Make Time You Are Here

Rodrigo Rivera is a machine learning researcher at the Advanced Data Analytics in Science and Engineering Group at Skoltech and technical director of Samsung Next. He's previously been in data science and research leadership roles at companies all around the world including Rocket Internet and Philip-Morris. Learn more about Rodrigo: https://rodrigo-rivera.com/ https://twitter.com/rodrigorivr Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: https://www.cyou.ai/newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 03:00 How Rodrigo got started in computer science and started his first company 10:40 Rodrigo's experiences leading data science teams at Rocket Internet and PMI 26:15 Leaving industry to get a PhD in machine learning 28:55 Data science collaboration between business and academia 32:45 Rodrigo's research interest in time series data 39:25 Topological data analysis 45:35 Framing effective research as a startup 48:15 Neural Prophet 01:04:10 The potential future of Julia for numerical computing 01:08:20 Most exciting opportunities for ML in industry 01:15:05 Rodrigo's advice for listeners 01:17:00 Rapid fire questions Links: Rodrigo's Google Scholar Advanced Data Analytics in Science and Engineering Group Neural Prophet M-Competitions Machine Learning Refined Foundations of Machine Learning A First Course in Machine Learning

Dan Jeffries is the chief technical evangelist at Pachyderm, a leading data science platform. He's a prominent writer and speaker on all things related to the future. He's been in software for over two decades, many of those at Redhat, and is the founder of the AI Infrastructure Alliance and Practical AI Ethics. Learn more about Dan: https://twitter.com/Dan_Jeffries1 https://medium.com/@dan.jeffries Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: http://cyou.ai/newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 02:15 How Dan got started in computer science 06:50 What Dan is most excited about in AI 14:45 Where we are in the adoption curve of ML 20:40 The "Canonical Stack" of ML 32:00 Dan's goal for the AI Infrastructure Alliance 40:55 "Problems that ML startups don't know they're going to have" 49:00 Closed vs open source tools in the Canonical Stack 01:00:05 Building out the "boring" part of the infrastructure to enable exciting applications 01:08:40 Dan's practical approach to AI Ethics 01:23:50 Rapid fire questions Links: Pachyderm AI Infrastructure Alliance Practical AI Ethics Alliance Rise of the Canonical Stack in Machine Learning Rise of AI - The Age of AI in 2030 Google Magenta AlphaGo Documentary Thinking in Bets A History of the World in 6 Glasses Super-Thinking

Willem Pienaar is the co-creator of Feast, the leading open source feature store, which he leads the development of as a tech lead at Tecton. Previously, he led the ML platform team at Gojek, a super-app in Southeast Asia. Learn more: https://twitter.com/willpienaar https://feast.dev/ Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: https://www.cyou.ai/newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 02:15 How Willem got started in computer science 03:40 Paying for college by starting an ISP 05:25 Willem's experience creating Gojek's ML platform 21:45 Issues faced that led to the creation of Feast 26:45 Lessons learned building Feast 33:45 Integrating Feast with data quality monitoring tools 40:10 What it looks like for a team to adopt Feast 44:20 Feast's current integrations and future roadmap 46:05 How a data scientist would use Feast when creating a model 49:40 How the feature store pattern handles DAGs of models 52:00 Priorities for a startup's data infrastructure 55:00 Integrating with Amundsen, Lyft's data catalog 57:15 The evolution of data and MLOps tool standards for interoperability 01:01:35 Other tools in the modern data stack 01:04:30 The interplay between open and closed source offerings Links: Feast's Github Gojek Data Science Blog Data Build Tool (DBT) Tensorflow Data Validation (TFDV) A State of Feast Google BigQuery Lyft Amundsen Cortex Kubeflow MLFlow

Benedikt Koller is a self-professed "Ops guy", having spent over 12 years working in roles such as DevOps engineer, platform engineer, and infrastructure tech lead at companies like Stylight and Talentry in addition to his own consultancy KEMB. He's recently dove head first into the world of ML, where he hopes to bring his extensive ops knowledge into the field as the co-founder of Maiot, the company behind ZenML, an open source MLOps framework. Learn more: https://zenml.io/ https://maiot.io/ Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: https://www.cyou.ai/newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 02:15 Introducing Benedikt Koller 05:30 What the "DevOps revolution" was 10:10 Bringing good Ops practices into ML projects 30:50 Pivoting from vehicle predictive analytics to open source ML tooling 34:35 Design decisions made in ZenML 39:20 Most common problems faced by applied ML teams 49:00 The importance of separating configurations from code 55:25 Resources Ben recommends for learning Ops 57:30 What to monitor in an ML pipelines 01:00:45 Why you should run experiments in automated pipelines 01:08:20 The essential components of an MLOps stack 01:10:25 Building an open source business and what's next for ZenML 01:20:20 Rapid fire questions Links: ZenML's GitHub Maiot Blog The Twelve Factor App 12 Factors of reproducible Machine Learning in production Seldon Pachyderm KubeFlow Something Deeply Hidden The Expanse Series The Three Body Problem Extreme Ownership

Josh Albrecht is the co-founder and CTO of Generally Intelligent, an independent research lab investigating the fundamentals of learning across humans and machines. Previously, he was the lead data architect at Addepar, CTO of CloudFab, and CTO of Sourceress, which Generally Intelligent is a pivot from. Learn more about Josh: http://joshalbrecht.com/ http://generallyintelligent.ai/ Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: https://www.cyou.ai/newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 02:15 Introducing Josh Albrecht 03:30 How Josh got started in computer science 06:35 Josh's first two startup attempts 09:15 The tech behind Sourceress, an AI recruiting platform 16:10 Pivoting from Sourceress to Generally Intelligent, an AI research lab 23:50 How Josh defines "general intelligence" 28:35 Why Josh thinks self-supervised learning is the current most promising research area 36:15 Generally Intelligent's immediate research roadmap: BYOL, simulated environments 59:20 How Josh thinks about creating an optimal research environment 01:11:35 The "why" behind starting an independent research lab 01:13:30 AI alignment 01:17:00 Rapid fire questions Links: Bootstrap your own latent: A new approach to self-supervised Learning Understanding self-supervised and contrastive learning with "Bootstrap Your Own Latent" (BYOL) BYOL works even without batch statistics Generally Intelligent Podcast Consequences of Misaligned AI Why We Sleep Peak

Elena Samuylova and Emeli Dral are the co-founders of Evidently AI, where they build open source tools to analyze and monitor machine learning models. Elena was previously the head of the startup ecosystem at Yandex, director of business development at their data factory and chief product officer at Mechanica AI. Emeli was previously a data scientist at Yandex, chief data scientist at the data factory and Mechanica AI in addition to teaching machine learning both online and at multiple universities. Learn more about Elena, Emeli, and Evidently AI: https://evidentlyai.com/ https://twitter.com/elenasamuylova https://twitter.com/EmeliDral Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: http://cyou.ai/newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 02:15 How Emeli and Elena each got started in data science 07:10 Applying machine learning across a wide variety of industries at the Yandex Data Factory 14:55 Using ML for industrial process improvement 23:35 Challenges encountered in industrial ML and technical solutions 27:15 The huge opportunity for ML in manufacturing 34:35 How to ensure safety when using models in physical systems 37:40 Why they started working on tools for data and ML monitoring 42:50 Different kinds of data drift and how to address them 48:25 Common mistakes ML teams make in monitoring 55:25 Features of Evidently AI's library 57:35 Building open source software 01:02:25 Technical roadmap for Evidently 01:05:50 Monitoring complex data 01:08:50 Business roadmap for Evidently 01:11:35 Rapid fire questions Links: Evidently on Github Evidently AI's Blog Thinking Fast and Slow Flow Doing Good Better

Harikrishna Narayanan is the co-founder of a YC-backed stealth startup. He was previously a Principal Engineer at Yahoo, a Director in Workday's Machine Learning organization, and holds an M.S. from Georgia Tech. Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: https://cyou.ai/newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 02:45 How Hari got started in computer science and machine learning 06:00 Making the transition from IC to manager 14:35 What it means to be an effective engineering manager 19:20 Differences in managing machine learning vs traditional software teams 24:30 The importance of explaining complicated topics simply 30:15 How he thinks about hiring for data science and machine learning 36:50 Mistakes Workday made as it adopted machine learning 41:50 Essential skills for machine learning engineers 54:05 Why the future of AI is augmentation, not automation 58:30 His experience so far with YC 01:02:00 Rapid fire questions Links: Growth Mindset The Feynman Technique Radical Candor Trillion Dollar Coach Multipliers Good to Great The First 90 Days Crossing the Chasm Zero to One The Lean Startup The Hard Thing About Hard Things Sapiens A Short History of Nearly Everything On Intelligence Prediction Machines Algorithms to Live By

Learn more about the ML Ops Community: https://mlops.community/ Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: https://cyou.ai/newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 02:45 Intro 04:10 How I got into data science and machine learning 08:25 My experience working as an ML engineer and starting the podcast 12:15 Project management methods for machine learning 20:50 ML job roles are trending towards more specialization 26:15 ML tools enable collaboration between roles and encode best practices 34:00 Data privacy, security, and provenance as first class considerations 39:30 The future of managed ML platforms and cloud providers 49:05 What I've learned about building a career in ML engineering 54:10 Dealing with information overload Links: Josh Tobin: Research at OpenAI, Full Stack Deep Learning, ML in Production The Third Wave Data Scientist Practical ML Ops // Noah Gift // MLOps Coffee Sessions Building a Post-Scarcity Future using Machine Learning with Pavle Jeremic (Aether Bio) SRE for ML Infra // Todd Underwood // MLOps Coffee Sessions Luigi Patruno on the ML Ops Community podcast Luigi Patruno: ML in Production, Adding Business Value with Data Science, "Code 2.0"

Pavle Jeremic is the founder and CEO of Aether Biomachines, one of the most exciting ML-powered startups I've come across. His mission is to solve scarcity and Aether is the first step towards that. He was recently featured in Forbes' 30 under 30 in Manufacturing and holds a B.S. in Biomolecular Engineering from UC Santa Cruz. Learn more: Aether Biomachines Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: http://bitly.com/mle-newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 02:45 Pavle Jeremic 05:20 How Pavle was introduced to computer science and programming 08:00 Solving scarcity from first principles 23:20 How Aether contributes to the post-scarcity future 29:30 What enzymatic reaction data looks like 37:20 Using deep learning to figure out what enzymatic experiments to run next 39:45 How Aether runs thousands of experiments at a time 47:00 What the current bottleneck of the system is 53:15 The evolution of ML models at Aether 59:00 Gaps in existing ML infrastructure solutions 01:03:30 Why Aether is releasing some of their data for a competition 01:06:50 The upcoming roadmap for Aether 01:09:30 Rapid fire questions Links: Founders First Interview - Making Alchemy Real DeepChem Engines of Creation Rama Series

Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: http://bitly.com/mle-newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 02:50 Josh Tobin: Research at OpenAI, Full Stack Deep Learning, ML in Production 21:48 Shreya Shankar: Lessons learned after a year of putting ML into production 34:44 Luigi Patruno: ML in Production, Adding Business Value with Data Science, "Code 2.0" 53:28 Music Information Retrieval at Spotify and the Future of ML Tooling with Andreas Jansson of Replicate

Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: http://bitly.com/mle-newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/

Andreas Jansson is the co-founder of Replicate, a version control tool for machine learning. He holds a PhD from City University of London in Music Informatics and was previously a machine learning engineer at Spotify, researching and applying algorithms for music information retrieval. Learn more about Andreas: https://replicate.ai/ https://www.linkedin.com/in/janssonandreas/ Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: http://bitly.com/mle-newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 02:30 Andreas Jansson 07:30 Overview of music information retrieval (MIR) 13:30 Why use spectrograms and not raw audio? 19:55 The potential for transformers in MIR 22:45 Most exciting applications for ML in MIR 29:20 Challenges in putting ML into production 36:45 What Andreas imagines for the future of ML tools 41:45 Why he's building a tool for ML version control (http://replicate.ai/) 52:55 What Replicate enables via integration or as a platform 01:02:55 Learnings from doing customer discovery for Replicate 01:14:10 "Github for ML models and data" 01:22:30 Rapid fire questions Links: WaveNet: a generative model for raw audio Singing Voice Separation with Deep U-Net CNNs Joint Singing Voice Separation and F0 Estimation with Deep U-Net Architectures arXiv Vanity Replicate Replicate's Discord

Luigi is the director of data science at 2U, where he leads a team in developing ML models and infrastructure to predict student success outcomes. He's also the founder of ML in Production, a blog and newsletter that helps readers build, deploy, and run ML systems. Learn more about Luigi: https://mlinproduction.com/ https://twitter.com/mlinproduction Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: http://bitly.com/mle-newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 02:45 Luigi Patruno 04:50 How can ML teams be more rigorous in their engineering practices? 10:25 Best practices for monitoring and logging ML systems 18:00 Adding business value with data science 37:10 Most valuable types of tools for ML in production 43:15 What an ideal data pipeline setup looks like 47:50 Unbundling the "Data Scientist" role 50:35 The future of building software: "Code 2.0" 59:45 Most valuable skills for the future 01:10:15 Learnings from writing his blog "ML in Production" 01:15:00 Rapid fire questions Links: Luigi's interview on Datacast Ultimate Guide to Deploying ML Models Maximizing Business Impact with Machine Learning Two Types of Companies Using ML The AI Hierarchy of Needs Josh Tobin: Research at OpenAI, Full Stack Deep Learning, ML in Production Machine Learning is Forcing Software Development to Evolve ML Street Talk #29: GPT-3, Prompt Engineering, Trading, AI Alignment, Intelligence Building Machine Learning Powered Applications How to Change Your Mind The War of Art

Shawn Wang formerly worked in finance as a derivatives trader and equity analyst before burning out and pivoting towards tech. He's a prolific blogger who goes under the pseudonym "swyx" and recently published the excellent Coding Career Handbook. He's a graduate of Free Code Camp and Full Stack Academy now working at AWS as a Senior Developer Advocate. Learn more about Shawn: https://swyx.io/ https://www.learninpublic.org/ https://twitter.com/swyx Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: http://bitly.com/mle-newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 02:45 swyx is back! 05:25 How his book has been received so far 11:35 Why and how to negotiate your salary 24:10 Getting started in public speaking, giving talks at meetups and conferences 35:45 The role of luck in your career and how to create it 51:15 Biggest is not best, best *for me *****is best 59:20 Why swyx angel-invested in Circle 01:12:00 On Randy Pausch's Time Management lecture 01:18:00 Using open source to accelerate your coding skill 01:20:00 Handling information overload and enhancing retention with note taking 01:27:20 What swyx does in his job as a Developer Advocate and why you should consider non-coding roles 01:37:30 swyx's new podcast Career Chats (https://careerchats.transistor.fm/) Links: swyx's first ML Engineered appearance swyx's book Coding Career Handbook How to Create Luck Notes on Time Management from a Dying Professor Building a Second Brain SimpleNote swyx's new podcast with Randall Kanna "Career Chats"

Yannic Kilcher is PhD candidate at ETH Zurich researching deep learning, structured learning, and optimization for large and high-dimensional data. He produces videos on his enormously popular Youtube channel breaking down recent ML papers. Follow Yannic on Twitter: https://twitter.com/ykilcher Check out Yannic's excellent Youtube channel: https://www.youtube.com/channel/UCZHmQk67mSJgfCCTn7xBfew Listen to the ML Street Talk podcast: https://podcasts.apple.com/us/podcast/machine-learning-street-talk/id1510472996 Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: http://bitly.com/mle-newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 02:40 Yannic Kilcher 07:05 Research for his PhD thesis and plans for the future 12:05 How he produces videos for his enormously popular Youtube channel 21:50 Yannic's research process: choosing what to read and how he reads for understanding 27:30 Why ML conference peer review is broken and what a better solution looks like 45:20 On the field's obsession with state of the art 48:30 Is deep learning is the future of AI? Is attention all you need? 56:10 Is AI overhyped right now? 01:01:00 Community Questions 01:13:30 Yannic flips the script and asks me about what I do 01:25:30 Rapid fire questions Links: Yannic's amazing Youtube Channel Yannic's Google Scholar Yannic's Community Discord Channel On the Measure of Intelligence: arXiv paper and Yannic's video series How I Read a Paper: Facebook's DETR (Video Tutorial) An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale (Paper Explained) Zero to One The Gulag Archipelago

Zak Slayback is a principal at 1517 Fund, a venture capital fund that prioritizes working with dropouts. He wrote the excellent book "How to Get Ahead", one of my most recommended books on careers, and runs Get Ahead Labs where he teaches how to write outstanding cold emails. Learn more about Zak: https://zakslayback.com/ https://www.1517fund.com/ Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: http://bitly.com/mle-newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bit.ly/mle-survey Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Timestamps: 02:35 Zak Slayback 04:45 Using opportunity cost, signaling theory, and incentives to accelerate your career (https://zakslayback.com/frameworks-success-opportunity-cost/) 14:35 How to set career goals (https://zakslayback.com/ambition-mapping/) 20:15 Rene Girard and Mimetic Desire 24:30 The difference between a mentor, a coach/consultant, and an advisor (https://zakslayback.com/whats-difference-mentors-advisors-coaches/) 35:40 Finding a mentor (https://zakslayback.com/professional-mentor-dream-job/) 44:30 Fighting mental blocks against reaching out to potential mentors 47:30 Why you should start a personal website (https://zakslayback.com/why-start-a-website/) 56:15 What the most important "meta-skills" are and how to stack talents 01:05:35 Most over-looked sections of the book 01:09:00 The future of higher education: the new 95 theses from 1517 Fund (https://medium.com/1517/a-new-95-ec071200d98f) 01:23:05 What Zak thinks the most exciting trends in technology are 01:35:15 Rapid fire questions Links: The End of School and Building a Valuable Skillset with Zak Slayback Deschool Yourself and Find Your Focus – With Zak Slayback Zak's book - How to Get Ahead (highly recommended!) Ambition Mapping Rene Girard and Mimetic Desire

Letitia Parcalabescu is a PhD candidate at the University of Heidelberg focused on multi-modal machine learning, specifically with vision and language. Learn more about Letitia: https://www.cl.uni-heidelberg.de/~parcalabescu/ https://www.youtube.com/channel/UCobqgqE4i5Kf7wrxRxhToQA Every Thursday I send out the most useful things I’ve learned, curated specifically for the busy machine learning engineer. Sign up here: http://bitly.com/mle-newsletter Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Subscribe to ML Engineered: https://mlengineered.com/listen Comments? Questions? Submit them here: http://bitly.com/mle-survey Timestamps: 01:30 Follow Charlie on Twitter (https://twitter.com/CharlieYouAI) 02:40 Letitia Parcalabescu 03:55 How she got started in CS and ML 07:20 What is multi-modal machine learning? (https://www.youtube.com/playlist?list=PLpZBeKTZRGPNKxoNaeMD9GViU_aH_HJab) 16:55 Most exciting use-cases for ML 20:45 The 5 stages of machine understanding (https://www.youtube.com/watch?v=-niprVHNrgI) 23:15 The future of multi-modal ML (GPT-50?) 27:00 The importance of communicating AI breakthroughs to the general public 37:40 Positive applications of the future “GPT-50” 43:35 Letitia’s CVPR paper on phrase grounding (https://openaccess.thecvf.com/content_CVPRW_2020/papers/w56/Parcalabescu_Exploring_Phrase_Grounding_Without_Training_Contextualisation_and_Extension_to_Text-Based_CVPRW_2020_paper.pdf) 53:15 ViLBERT: is attention all you need in multi-modal ML? (https://arxiv.org/abs/1908.02265) 57:00 Preventing “modality dominance” 01:03:25 How she keeps up in such a fast-moving field 01:10:50 Why she started her AI Coffee Break YouTube Channel (https://www.youtube.com/c/AICoffeeBreakwithLetitiaParcalabescu/) 01:18:10 Rapid fire questions Links: AI Coffee Break Youtube Channel Exploring Phrase Grounding without Training

Moin Nadeem is a masters student at MIT, where he studies natural language generation. His research interests broadly include natural language processing, information retrieval, and software systems for machine learning. Learn more about Moin: https://moinnadeem.com/ https://twitter.com/moinnadeem Want to level-up your skills in machine learning and software engineering? Join the ML Engineered Newsletter: http://bit.ly/mle-newsletter Comments? Questions? Submit them here: http://bit.ly/mle-survey Follow Charlie on Twitter: https://twitter.com/CharlieYouAI Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ Subscribe to ML Engineered: https://mlengineered.com/listen Timestamps: 01:35 Follow Charlie on Twitter (https://twitter.com/CharlieYouAI) 03:10 How Moin got started in computer science 05:50 Using ML to identify depression on Twitter in high school 11:00 Building a system to track phone locations on MIT’s campus 14:35 Specializing in NLP 17:20 Building an end-to-end fact-checking system (https://www.aclweb.org/anthology/N19-4014/) 25:15 Predicting statement stance with neural multi-task learning (https://www.aclweb.org/anthology/D19-6603/) 27:20 Is feature engineering in NLP dead? 29:40 Reconciling language models with existing knowledge graphs 35:20 How advances in AI hardware will affect NLP research (crazy!) 47:25 Moin’s research into sampling algorithms for natural language generation (https://arxiv.org/abs/2009.07243) 57:10 Under-rated areas of ML research 01:00:10 How research works at MIT CSAIL 01:04:35 How Moin keeps up in such a fast-moving field 01:11:30 Starting the MIT Machine Intelligence Community 01:16:30 Rapid Fire Questions Links: FAKTA: An Automatic End-to-End Fact Checking System StereoSet: Measuring stereotypical bias in pretrained language models Neural Multi-Task Learning for Stance Prediction Rich Sutton - The Bitter Lesson A Systematic Characterization of Sampling Algorithms for Open-ended Language Generation Strategies for Pre-training Graph Neural Networks Transformers For Image Recognition at Scale