The Minhaaj's Podcast

Ankit is an experienced AI Researcher/Machine Learning Engineer who is passionate about using AI to build scalable machine learning products. In his 10 years of AI career, he has researched and deployed several state-of-the-art machine learning models which have impacted 100s of millions of users.   

Currently, He works as a senior research scientist at Facebook where he works on a variety of machine learning problems across different verticals. Previously, he was a  researcher at Uber AI  where he worked on application of deep learning methods to different problems ranging from food delivery, fraud detection to self-driving cars.   He has been a featured speaker in many of the top AI conferences and universities like UC Berkeley, IIT Bombay and has published papers in several top conferences like Neurips, ICLR. Additionally, he has co-authored a book on machine learning titled TensorFlow Machine Learning Projects.  He has undergraduate and graduate degrees from IIT Bombay (India) and UC Berkeley respectively. Outside of work, he enjoys running and has run several marathons.

00:00 Intro

00:17  IIT vs FAANG companies, Competition Anxiety

05:40  Work Load between India and US, Educational Culture  

07:50. Uber Eats, Food Recommendation Systems and Graph Networks 

11:00 Accuracy Matrices for Recommendation Systems  

12:42 Weather as a predictor of Food Orders and Pizza Fad

15:48 Raquel Urtusun and Zoubin Gharamani, Autonomous Driving and Google Brain

17:30 Graph Learning in Computer Vision & Beating the Benchmarks

19:15 Latent Space Representations and Fraud Detection

21:30 Multimodal Data & Prediction Accuracy 

23:20 Multimodal Graph Recommendation at Uber Eats

23:50 Post-Order Data Analysis for Uber Eats

27:30  Plugging out of Matrix and Marathon Running

31:44  Finding Collusion between Riders and Drivers with Graph Learning 

35:40  Reward Sensitivity Analysis for Drivers in Uber through LSTM Networks 

42:00 PyG 2.0, Jure Leskovec, and DeepGraph, Tensorflow Support  

46:46 Pytorch vs Tensorflow, Scalability and ease of use.

52:10 Work at Facebook, End to End Experiments

55:19 Optimisation of Cross-functional Solutions for Multiple Teams  

57:30  Content Understanding teams and Behaviour Prediction

59:50 Cold Start Problem and Representation Mapping 

01:03:30 NeurIPS paper on Meta-Learning and Global Few-Shot Model

01:07:00 Experimentation Ambience at Facebook, Privacy and Data Mine 

01:09:03 Cons of working at FAANG 

01:10:20 High School Math Teacher as Inspiration and Mentoring Others 

01:18:25 TensorFlow Book and Upcoming Blog

01:16:40 Working at Oil Rig in the Ocean Straight Out of College 

01:20:08 Promises of AI and Benefits to Society at Large

01:25:50 Facebook accused of Polarisation, Manipulation and Racism 

01:28:10 Revenue Models - Product vs Advertising

01:31:15 Metaverse and Long-term Goals 

01:33:10 Facebook Ray-Ban Stories and Market for Smart Glasses

01:36:40 Possibility of Facebook OS for Facebook Hardware

01:38:00 LibraCoin & Moving Fast - Breaking Things at Facebook 

01:39:09 Orkut vs Facebook - A case study on Superior Tech Stack

01:42:00 Careers in Data Science & How to Get into It

01:45:00 Irrelevance of College Degrees and Prestigious Universities as Pre-requisites

01:49:50 Decreasing Attention Span & Lack of Curiosity 

01:54:40 Arranged Marriages & Shifting Relationship Trends

Francis Corrigan is Director of Decision Intelligence at Target Corporation. Embedded within the Global Supply Chain, Decision Intelligence combines data science with model thinking to help decision-makers solve problems.

00:00 Intro 

01:21  Data Science applications in Logistics and Supply Chain, Cost and Performance trade-off 

03:21  Amazon vs Target fulfillment Model, Owning vs Coordinating with Last Mile companies e.g. FedEx 

08:36 Suez Canal Container Blockage, Fallback plan at Target 

10:37 Predicting products to Stock in Bottle Neck Scenarios 

12:42  Air Freight vs Sea Shipments Costs, Ideal vs Real World Deliveries 

15:48  Lack of Good Data and Prediction Challenges 

18:00  Managing Expectations as Head of Analytics, Importance of Communicating 

20:11 Stakeholder Management & Data Science  Newsletter 

23:39  Technical and Non-technical Teams Coordination, Speed Reading 

26:36  Data Stories and Visualizations 

29:47 Reporting Pipelines vs Story Narration 

31:37 Times Series, Prophet, Flourish and Hans Rosling 

35:28  Economist turned Data Scientist,  Embarrassment as Motivation 

38:20  Lack of Practical Skills of Data Science at University 

41:18  Employer’s Perspectives on Data Science Talent  

45:24  What Causes Data Teams Failure 

48:40  COVID 19 and Times Series Corruption,  Anomaly Detection 

56:15  Toilet Paper Demand Scenario, Commodity Pricing Alerts 

59:50  Automating Alerts for Panic Situation 

01:02:10  Pandemic as a Blessing for Digital Business, Exponential Growth Rates and Tuition Fee Reimbursement for Employees 

01:06:06 Data as Decision Support System, Strategic Decision Indicators  

01:08:08 Capital in 21st Century, Thomas Piketty and Free Markets  

01:11:31 Failures of Capitalist Societies on Individual Front and Socialist Aversion of Wealth Generation  

01:15:15 UBI, Interventions, and CEO to Lowest Paid Worker Ratio  

01:18:25 Career Blunders and Regrets  

01:22:12 Psychometric Tests for Intellect Filtering, Behavioral Stability and Creativity Trade-off  

01:24:08 Target’s Epic Failure in Canada, What Data Science could have Prevented  

01:25:08 Gameplan to Compete with Walmart and Amazon   

01:28:00 Sarimax, Armiax and Volatility Management, Planning vs Forecasting 

01:31:33 Deep NNs or Lack thereof, Explainability and Monte Carlo as Alternative 

01:34:00 Model Parsimony in Times Series, Baseline Models in Excel  

01:37:50 R vs Python, Specific Use Cases  

01:40:25 Delegating and Element of Trust  

01:43:20 Time and Space Complexity of Models, Netflix and Deployments at Target  

01:46:00 Political Impacts on Shipments, Narratives and Hypothesis Testing 

01:48:00 Nate Silver, Nassem Talib, and Early Inspirations 

01:52:05 Work-life Balance

Walid S. Saba is the Founder and Principal AI Scientist at ONTOLOGIK.AI where he works on the development of Conversational AI. Prior to this, he was a PrincipalAI Scientist at Astound.ai and Co-Founder and the CTO of Klangoo. He also held various positions at such places as the American Institutes for Research, AT&TBell Labs, Metlife, IBM and Cognos, and has spent 7 years in academia where he taught computer science at the New Jersey Institute of Technology, theUniversity of Windsor, and the American University of Beirut (AUB). Dr. Saba is frequently an invited speaker at various organizations and is also frequently invited to various panels and podcasts that discuss issues related to AI and Natural Language Processing. He has published over 40 technical articles, including an award-winning paper that was presented at theGerman Artificial Intelligence Conference in 2008. Walid holds a BSc and an MSc in Computer Science as well as a Ph.D. in Computer Science (AI/NLP) which he obtained from Carleton University in 1999.

00:00 intro

01:00 Language as a mental construct, PAC,  Subtext in Sentences

06:28 OpenAI’s Codex Platform, Below Human Baseline Performance of NLP

18:00 Comprehension vs Generation,  Search vs Context

19:20 Sophia the Robot, Shallow ethics in AI and Commercialisation of Academia

27:40 Bad Research Papers, Facebook runaway train & AI Godfathers Cult.

32:30 AI leaders and Profiteering, Unethical Behaviour of Influencers.

37:50 Non-Verbal Component of Natural Language Understanding, Prosody and Accuracy Boost

41:33 Ontologik’s NLU Engine, Adjective Ordering Restriction Mystery 

43:58 Ontological Structure and Chomsky’s Universal Grammar, Discovery vs Creation

45:31 Entity Extraction and How Ontologik’s Engine tackles this Problem

47:50 Language Agnostic Learning, Foreign Language Learning, and Pedagogy of Linguistics

54:00 First Language, Blank State and Missing Sounds in Some Languages

55:20 Real-time Language Translation Engines, AR/VR Aids and Commercial Utility

01:01:00 Sentiment Analysis, Language Policing & Censorship 

01:04:00 Ontological Structures, Gender Bias and Situational Paradox

01:09:00 3 Foods for Rest of the Life & Fad Food Indulgence

01:11:00 Inspiration for Getting into the Field, Career Ideals & Cultural Influence

01:15:30 Epistemology, IQ and The Bell Curve 

01:17:00 Einstein’s IQ, Haircut, Social Skills, and Success Rubric

01:22:00 Attracting Brilliant Talent Around the World, Ivy League PhDs & Standardised Testing

01:28:40 Unsupervised Learning, Accuracy & Comprehensibility in NLU

01:30:20 BF Skinner, Pavlovian Dogs, Skinner has been Skinned.

01:37:50 Human Behavioral Biology, Endocrinal System similarities with Humans yet they don’t learn Languages.

01:45:30 Language as an expression of Genetic differences, Big Five & Phenotype.

01:49:40 IBM Watson Personality Insights, Text-based personality Inferences.

01:55:30 Long Short Term Memory Issue in Ontologik’s Engine, Computational Complexity, Timeline for Release

Richard Turrin is an award-winning, dynamic Fintech expert with over 20 years of experience in leveraging new technology to drive revenue growth for market-leading companies. He writes, speaks, and consults on innovation and China because he learned about both the hard way. He headed four labs in banks responsible for financial product innovation and headed fintech operations in China. His best-selling books 'Cashless' and 'Innovation Excellence Lab' are some of the most recommended books on the topic. He has worked for IBM in senior roles and has been a professor at Hult International Business School. 

00:00 intro

00:38 Financial Crisis of 90s and Arrival in Shangai

03:32 China, From Copycat to Innovator, WeChat and Mark Zuckerberg

08:42 Twelfth Five-Year Plan, Regulation of IT and China’s Gorbachev of Fintech

15:09 Crocodile in the Yangtze, Chinese’s Public-Private Partnership and Liberalisation

24:19 China’s Great Firewall, Invasion of Privacy and Uhygar Plight, Freedom as an Average Consumer

32:24 Central Bank Digital Currency, Blockchain and Version 2.0

42:00  Architecture of CDBC. Centralised, Decentralised or Distributed. Public or Private BlockChain

54:00  Challenges of International Adoption of China’s CDBC , US’s Exclusionary Policy and China’s Partnerships in Africa and Middle East 

01:11:50 Losers in Chinese CDBC War, US and Allies, PayPal, Square, Stripe and Alipay

01:29:30 US Foreign Policy, Walk of Shame out of Afghanistan, Fiascos in Iraq and Vietnam and Lost Respect in International Community

01:39:50 China’s Innovation beats US by at least 5 Years, Repercussions of CDBC, Smart Contracts and Private Sector Trade

01:12:21 Inspiration in Life, Love of Books and Loss of Mother

James Gaskin is a professor of Information Systems Management at Brigham Young University.

00:00 intro

07:00 Mismatch between Pedagogical Style and Student Needs 

12:11 Behavioral Science and Structural Equation Modelling.

16:40  Cut-off scores in Model Acceptance in SEM

17:30  SEM on Big Data and Computational Intensity

20:14  Model Explanation in Academics vs Industry

21:16  Academic Trash Research Papers getting Published

23:35  Gerry-mandering of Data and Publishing Mafia

26:20  HTMT as Discriminant Validity Criteria and other Accuracy Metrics

28:20  Empirical vs Mathematical Convergence of Models and Compromise

29:51 Life as Gymnast and 31 Moves in Life, Japan, California and Malaysia

35:00 Failing High School and Relationship Stability 

37:19 Data Scientist vs Academic Learning - Siloed Knowledge

49:39 Book ‘How to lie with Statistics’, John Perkins, Joseph Stiglitz and Political Manipulation of Data 

44:22 Enron, Big Four and Cooking Books 

47:26 Hedonist Motivation System, Pavlov’s Reinforcement and SEM

53:40 Futility of Research in Social Sciences and Remedy

55:40  Utrect University Abandons Citations for Hiring and Promotion of Faculty for Open Science

59:40 Work Path towards Ph.D in Australia. Practice vs Theory

01:03:00 Fixing the Academia and Failure. Video Journals for Research

01:07:49 Ratemyprofessor Score and Wrong Incentives for Teacher Rating

01:10:00 University Campus as Political Battlefields, Psychology and Conservatism

01:12:21 Dilemma of Academia vs Industry in Future

01:31:51 Favourite SEM software

01:17:40 Shortcomings of AMOS by IBM 

01:19:36 Mediation and Moderation in Structural Models and Explainability

01:23:20 Dropout Regularisation, Second Generation Statistical Tools & Explainability

01:26:00 HC Moneyball and SEM for Understanding Dynamics

01:31:00  Path Analysis and Sales Data Modelling

01:532:41 Mediation and Moderation analysis and Prediction 

01:37:40 The Invention Book and Mechanical Engineer

01:40:15  Daughters, Invention Book and Different temperaments

01:53:41 Mediation and Moderation analysis and Prediction 

This week I'll sit down with Ganes Kesari, Chief Decision Scientist at Gramener, the company behind Gramex and an innovative rising star. Ganes is a contributor in leading magazines such as Forbes, TechCrunch, Entrepreneur, and The Enterprisers Project.  He won the 2020 CSuite Award for best blog by a business leader. He has taught at Princeton University and Indian Business School and has been invited to speak at TED, O'Reilly Strata, Microsoft, and Intel events.

Timestamps

00:00 intro

03:04 Gaining Business Value from AI Initiative,  Establishing Baselines  

08:41 Human vs AI performance baseline, Long Term Benefits of AI

10:30  Netflix’s Recommendation Engine Competition, Failed ROI and Model Accuracy

15:40  GCP losses, Data and Model Drifts,  Bulls Eye in a Moving Target

19:40 Data Maturity Assessment Tool,  5 Stage Roadmap 

23:20  Logging in Personal Journal, 4 years, 120,000 data points.

25:40  From Siloed Modules to End to End Flow, Gramex Unified Architecture

29:30 Gramener Game Plan, Services vs Platform

34:28 Using ModelHandler to Furnish Realtime Business Visualizations

36:51 WorldBank Data Visualisation of Technology and Entrepreneurship Report, Data Story Component

42:33 Data Journalism at Guardian,  Character and Plot Visualisation of Hindu Epic Saga Mahabharata, Shakespeare’s Sonnets, Hans Rosling and GapMinder

47:18 Airtel Contract Deals

49:51 IITs, IIMs, Geeks and Humor

54:22 Education in India, SAT scores and Path Ahead

58:50  Industry-Academic Partnerships and Practical Experience

01:03:00 AI Adoption Pain Points, Cybersecurity, Regulation, Fairness and Explainability 

01:07:00  CNA Financial $40M ransom payment and AI Adoption Correlation

01:13:00 Increasing Bain & Company’s Net Promote Score for a Computer Manufacturer

01:20:19 Backing in Himalayas and Monastery in Bhutan

01:25:25 AI in Biodiversity, Rhinoceros, Penguins and Whale Shark as Endangered Species

01:30:01  Google Vertex AI, Alteryx, Knime vs Gramex, Future Strategy

01:35:30 Slidesense, Business Reports and Powerpoint Integration

01:39:26 Explaining DeepLearning to your Daughter, Whitehat Jr Scam

01:45:00 How Technology is changing Social Landscape

01:47:26 Chess Champion, Garry Kasparov and DeepBlue Game

01:51:30  Chess and IQ, Narrow Intelligence and Transferability  

01:53:01  Tesla Killing Jaywalker and AI’s Mindless Application

01:55:00  G7 Summit 2020 and AI war between US & China

01:58:33 Smart Twins, Enterprise Mass Production & Gramex

Serg Masis is the author of best-selling book 'Interpretable Machine Learning with Python' and senior Data Scientist at Sygenta. He has mentored many data scientists around the world. 

Timestamps:

00:00 intro

08:30  Old 4.77 MH  z Computer, Late 80s and Programming

11:51 Fairness, Accountability and Transparency in Machine Learning, Startup and Harvard

16:33  Fairness vs Preciseness, Bias and Variance Tradeoff, Are Engineers to blame?

21:43 Mask-Detection Problem in Coded-Bias, Biased Samples,  Surveillance using CV

32:38 Fixing Biased Datasets, Augmenting Data and Limitations 

37:39 Algorithmic Optimisation and Explainability

40:51 Eric Schmidt on Behavioral Prediction, SHAP values, Tree and DeepExplainers

44:50 Challenges of using SHAP and LIME & Big Data

49:37 GPT3, Large Models and ROI on Explainability

01:00:00  TCAS, Collision Risks and Interpretability, Ransom Attacks

01:08:09 Guitar, Bass, and Led Zepplin

01:09:31 Birth Order and IQ, Science vs Folk Wisdom

01:13:30  Reverse Discrimination & Men, Bias in Child Custody, Prison Sentences, and Incarceration

01:23:11 Receidivism to Criminal Behaviour, Ethnic over-representation & Systematic Racism

01:24:44  Human Judges vs AI,  Absolute Fairness, Food and Parole

01:30:20  Face Detection in China, Privacy vs Convenience, Feature Engineering and Model Parsimony 

01:35:51 Sparsity, Interaction Effects, and Multicollinearity

01:38:23  Four levels of Global and Local Predictive Explainability

01:43:17  Recursive and Sequential Feature Selection

01:47:42  Ensemble, Blended and Stacked Models and Interpretability

01:53:45  In-Processing and Post-Processing Bias Mitigation

01:57:00  Future of Interpretable AI

Ryne Sherman is Chief Science Officer of Hogans Assessments and Podcast Host of Hogan's Podcast. He was a Professor of Psychological Sciences at Florida Atlantic University before that.   

TimeStamps:  

00:00​ intro 

01:42​ Personality Portraits,  Big Five, Self vs Other Reports 

03:53​  Trait Identification, Talent Hunting and Reputation-based Prediction 

06:07​  Adult Variations in Personality, High School Screening and Stability of Profile 

08:03​  Personality Profile of Trump Supporters and Shard Psychographics 

11:40​  Core Beliefs and Values as Predictors,  Politics and Economic Stimulus 

16:01​ Psychometric Theory vs Pop Psychology, Response Patterns and Behavioral Predictions 

19:11​  From Archetypes to Oedipus complex, Reliability and Validity of Hogan Assessments 

22:37​  Machine Learning, AI and Personality Predictions, Ensemble Models 

26:11​  Algorithmic Explainability in Assessment Data,  Avoiding Blackbox NNs 

29:02​ Career Development, Executive Recruiting and  Personality Plasticity 

32:00​  Department transfers, Perceived Image and Reputation Awareness 

33:51​ Childhood, Boy Scout and High School 

35:24​  Birth Order and Research on Effects on Personality 

37:00​  Dark Side of Personality, Attention Craving and Workplace Problems 

39:52​  Explaining Personality Reports and Failed Predictions

Greg Gage is the co-founder and CEO of Backyard Brains, an organization that develops open-source tools that allow amateurs and students to participate in neural discovery.  Greg is an NIH-award-winning neuroscientist with 9 popular TED Talks and dozens of peer-reviewed publications. Greg is a Senior Fellow at TED and the recipient of the White House Champion of Change from Barack Obama award for his commitment to citizen science.  

Greg Gage  

00:00 intro 

02:28 Graduate Work to Brain Interface Company 20:40 Neuralink, EMG and Cyborgs 

28:57 Electrode Scarring, Heart Stunts and Neural Engineering 

09:40  Neuronal Activation for Behavioral Activations in Monkeys 

38:39 Behaviorism, Experimental Psychology and Implications 

40:41 Recreation of Memories through Artificial Hippocampus 

41:40  Neural Network-based Prosthetics for Limb Amputees 

46:50  AI bot Sofia, Facial Nerves in Robots for Emotive Abilities 

51:44  Big Five, Personality Traits, Neurophysiology & Predicting Divorce 

57:41 Mental Disorders and Wearable Tech  

01:01:04 Surveys, Behavioral Data and Neuroscience 

01:04:00  Neuroscience in Schools and Expansion to Developing World

01:16:00 Work with LEXUS designing Autonomous Car Experience, Children and Science 

01:12:30  Community Work, Silicon Valley and Work Culture 01:20:00 RoboRoach, Flint Michigan and  Joy of Learning

Fabio Pereira is leading the Open Innovation Labs initiative in Latin America at Red Hat. Open Innovation Labs is an intensive, highly focused residency in an environment designed to experiment, immerse and catalyze innovation. He is the author of the book "Digital Nudge: The hidden forces behind the 35,000 decisions we make every day".   He had been a Principal Consultant, a Digital Transformation Advisor at ThoughtWorks, a large software consultancy firm for over 10 years.

Fabio Pereira

00:00 intro

00:30  35,000 Decisions , Cognitive Overload and Delegation to Technology

02:30 Pre-Modern Man’s Cognitive Load and Number of Choices 

13:52  Antidote to Decision Fatigue, Pomodoro Technique and CrossFit

09:40 Digital Quotient (DQ), Emotional Quotient (EQ) and IQ

14:10 Digital Nudge, Behavioral Sciences and Netflix

20:41 ThoughtWorks, Business Process and Insurance Company Use Case

26:20  UX and RX  in Book Writing, Summarization and Legibility

30:20 Nir Eyal, Indestructible and Divided Attention

31:28 Growing up in Brazil, Movies with Subtitles and Hobbies

35:45 IRC, Global Citizenship, Pharmacy and World Travel

44:53 Snapstreak, YouTube Videos, Notifications, Dopamine and Loss Aversion

48:26 John Suler’s 6 Factors of Inhibition,  Dual Identity and Twitter vs Linkedin

55:00  Clubhouse, Leaked Recordings and Real Self

56:00  Big Brother Brazil, Reality Shows and Evocation of Real Emotions

57:55 Cyborgs, Biohacking, Neuralink teaches Monkey’s to Play Games and Earthquake in Chile

01:01:00 Induced Emotions through Neurotransmitters and Hunger Aggression

01:05:00  Human Behaviour Biology, Robert Sapolsky and Diet/Memory Relationship

01:07:00 Thinking Fast and Slow and Ivy League baffling Bat and Ball Riddle

01:12:30 Intuition, ESP, Meditation and Eckhardt Tolle

01:18:12 Playing Prank on TEDx Audience, GDPR cookies and Privacy Policy Agreement

01:22:00 Default Biases for Visitors, Checkboxes, Radioboxes and Automatic Suggestions 

01:29:52 Algorithmic Bias, Newzealand’s AI-based Passport issuance and Movie Coded-Bias

01:38:15 Intentional Bias, Diversified Training Set and Double-Edged Sword

01:41:40 Moral Decisions for Self-Driving Car, MIT Review article on flawed IMAGENET  data

01:49:25 Time Well Spent movement , AR/VR tools for patients in Hospitals and Digital Nudging Tools

01:56:20  From CBT to Self-Assessing Behavioral Patterns

01:57:30  Innovation and Work at RedHat and Infobizity

02:01:00  Steve Wozniak, CS101 and Goals for ‘Digital Nudge’