The Data Scientist Show

Vikram is the co-founder of Galileo – an AI diagnostics and explainability platform used by data science teams building NLP, LLMs and Computer Vision models across the Fortune 500 and high growth startups. 
 Prior to Galileo, Vikram led Product Management at Google AI, where his team built models for the Fortune 2000 across retail, financial services, healthcare and contact centers. He has a master degree from Carnegie Mellon University from the school of computer science. If you enjoy the show, subscribe to the channel and leave a 5-star review. Subscribe to Daliana's newsletter on www.dalianaliu.com for more on data science and career.

Resources:LLM Studio: https://www.rungalileo.io/blog/announcing-llm-studio

Galileo: https://www.rungalileo.io/

Blog on LLM Hallucination: https://thesequence.substack.com/p/guest-post-stop-hallucinations-from

Vikram Chatterji’s LinkedIn: https://www.linkedin.com/in/vikram-chatterji/

"The Mom Test": https://www.amazon.com/The-Mom-Test-Rob-Fitzpatrick-audiobook/dp/B07RJZKZ7F

Daliana's Twitter: https://twitter.com/DalianaLiu

Daliana's LinkedIn: https://www.linkedin.com/in/dalianaliu

(00:00:00) Introduction

(00:04:24) How he got into machine learning

(00:06:53) Diagnosing large language models

(00:09:56) Addressing model hallucination

(00:12:46) Metrics for measuring hallucination

(00:17:30) From Google AI to starting Galileo

(00:24:08) Developing LLMs and putting them into production

(00:32:51) Galileo's diagnostics and explainability platform

(00:43:16)  Advice for data scientists when joining a startup

Max Pumperla designed his own career path in data science. He is a freelance software engineer at AnyScale, and also a data science professor. We talked about reinforcement learning, open source contributions, Ray for data scientists, and his view on the data scientists role. If you enjoy the show, subscribe to the channel and leave a 5-star review. Subscribe to Daliana's newsletter on www.dalianaliu.com for more on data science and career.

Max’s LinkedIn: https://www.linkedin.com/in/max-pumperla-a8099354/

Max's GitHub: https://github.com/maxpumperla

Daliana's Twitter: https://twitter.com/DalianaLiu

Daliana's LinkedIn: https://www.linkedin.com/in/dalianaliu

(00:00:00) Introduction

(00:09:19) How he got a remote job through Twitter

(00:14:06) Introduction to Ray

(00:18:52) Reinforcement learning

(00:23:56) Key lessons on integrating customer feedback

(00:35:12) Flaws in data science job titles

(00:45:51) How to be irreplaceable as a data scientist

(00:48:55) An unconventional career path as a data scientist

(01:12:24) Productivity and work-life balance

(01:28:10) Advice for building a personal brand

Piero Molino was one of the founding members of Uber AI Labs. He worked on several deployed ML systems, including an NLP model for Customer Support, and the Uber Eats Recommender System. He is the author of Ludwig , an open source declarative deep learning framework. In 2021 he co-founded Predibase, the low-code declarative machine learning platform built on top of Ludwig.Piero's LinkedIn: https://www.linkedin.com/in/pieromolino

Predibase free access: bit.ly/3PCeqqw

Daliana's Twitter: https://twitter.com/DalianaLiu

Daliana's LinkedIn: https://www.linkedin.com/in/dalianaliu

(00:00:00) Introduction

(00:01:54) Journey to machine learning

(00:03:51) Recommending system at Uber Eats

(00:04:13) Projects at Uber AI 

(00:09:34) Uber's customer obsession ticket system

(00:16:01) How to evaluate online-offline business and model performance metrics

(00:17:16) Customer Satisfaction

(00:28:38) When do you know whether a project is good enough

(00:41:50) Declarative machine learning and Ludwig

(00:45:32) Ludwig vs AutoML

(00:54:44) Working with Professor Chris Re

(00:58:32) Why he started Predibase

(01:07:56) LLM and GenAI

(01:10:17) Challenges for LLMs

(01:22:36) Advice for data scientists

(01:34:29) Career advice to his younger self

Holger Teichgraeber is a Data Science Manager at Archer Aviation. Previously, he worked at Convoy as a Research Scientist on their trucking marketplace, and at various companies in the energy space. Holger has a Bachelor's degree in Mechanical Engineering from Aachen, Germany, and a Masters and Ph.D. with research focus on machine learning and optimization applied to energy systems from Stanford University. He regularly writes on LinkedIn, with the goal to show how to build valuable products at the intersection of machine learning and optimization in production. If you enjoy the show, subscribe to the channel and leave a 5-star review. Subscribe to Daliana's newsletter on www.dalianaliu.com for more on data science and career.

Holger's LinkedIn: https://www.linkedin.com/in/holgerteichgraeber/

Daliana's Twitter: https://twitter.com/DalianaLiu

Daliana's LinkedIn: https://www.linkedin.com/in/dalianaliu

(00:00:00) Introduction

(00:01:28) How he got into operations research

(00:02:39) Operation research vs data science

(00:04:37) Trucking optimization at Convoy

(00:08:42) Optimization problem

(00:10:18) Strategic planning on air mobility at Archer

(00:13:50) Using simulation and solving a problem

(00:16:45) Big data science work vs smaller data science work

(00:21:23) Stakeholder management

(00:29:28) IC vs Manager

(00:32:04) Advice on promotion

(00:39:12) Work cultures in Germany and the US

(00:41:16) How to handle tight deadlines

(00:43:21) Important feedback from his work

(00:44:14) How to plan projects

(00:44:45) Next big challenge for data science teams

(00:45:40) Career growth in the next few years

(00:46:01) Connect with Holger

Barr Moses is a consultant turned CEO & Co-Founder of Monte Carlo, a data reliability company. She started her career as a management consultant at Bain & Company and a research assistant at the Statistics Department at Stanford University. Later, she became VP of Customer Operations at customer success company Gainsight, where she built the data and analytics team. She also served in the Israeli Air Force as a commander of an intelligence data analyst unit. Barr graduated from Stanford with a B.Sc. in Mathematical and Computational Science. Today, we’ll talk about Barr’s career journey, data reliability and observability, and what it means for data teams. If you enjoy the show, subscribe to the channel and leave a 5-star review. Subscribe to Daliana's newsletter on www.dalianaliu.com for more on data science.Barr's LinkedIn: https://www.linkedin.com/in/barrmoses/Daliana's Twitter: https://twitter.com/DalianaLiuDaliana's LinkedIn: https://www.linkedin.com/in/dalianaliu

(00:00:00) Introduction

(00:01:24) How did she got into data science

(00:08:26) Frameworks for data-driven decisions

(00:11:20) Is customer support ticket always bad?

(00:15:20) How to quickly find out what is true

(00:20:17) Struggles in the data team

(00:23:37) Daliana’s story about lineage

(00:28:00) People stressed about data

(00:28:09) Netflix was down because of wrong data

(00:30:40) Common issues with data quality

(00:33:14) 5 pillars of data observability

(00:39:14) How does Monte Carlo help data scientists

(00:43:08) Build in-house vs adopt tools

(00:45:48) How Daliana fixed a data quality issue

(01:02:44) How to measure the impact of the data team

(01:09:09) Mistakes she made

(01:15:28) Beat the odds

Deedy Das is a founding engineer at Glean, an enterprise search startup. Previously, he was a Tech Lead at Google Search working on query understanding and the sports product in New York, Tel Aviv, and Bangalore. Before that, he was an engineer at Facebook New York and graduated from Cornell University. Outside of work, Deedy writes on his blog. He published a viral resume template and his work on exposing grading flaws in the Indian education system. He also enjoys running marathons, road cycling, and playing cricket. Today we’ll talk about the search projects he worked on at Google, why he left Google, his current work at Glean, and his thoughts on whether Google is doomed because of  ChatGPT. If you enjoy the show, subscribe to the channel and leave a 5-star review. Subscribe to Daliana's newsletter on www.dalianaliu.com for more on data science.  

Deedy's Twitter: https://twitter.com/debarghya_das?s=20

Daliana's Twitter: https://twitter.com/DalianaLiu

Daliana's LinkedIn: https://www.linkedin.com/in/dalianaliu 

(00:00:00) Introduction 

(00:01:52) What is search 

(00:04:33) Query understanding 

(00:12:46) Google vs ChatGPT 

(00:18:24) Fixing bug for Sundar Pichai 

(00:27:33) Why he left google 

(00:30:32) How to get into search 

(00:34:38) Enterprise search at Glean 

(00:46:55) Advice for people who got laid off 

(00:48:41) What do search engineers do 

(00:51:37) How he evaluates candidates 

(00:53:58) Future of search 

(00:57:16) Why the web is declining 

(00:59:25) Copilot and AI-powered developer tools 

(01:03:46) Indian startup ecosystem 

(01:07:45) India vs Silicon Valley 

(01:09:48) How he grew 30k followers on Twitter 

(01:13:28) Daliana and Deedy’s challenge with social media 

(01:19:31) Career mistakes he made

Jeremy Nixon is a machine learning researcher, software engineer, and startup founder. Previously he was a software engineer at Google Brain working on deep learning. Now, he is the co-founder and CEO of Omni, building an immersive information retrieval system for you and your team. He studied applied math at Harvard University. Today we’ll talk about how he got into Google brain, his 3-month self-learning plan to learn machine learning, his startup, and how he executed his goal relentlessly since 2016. If you enjoy the show, subscribe to the channel and leave a 5-star review. Subscribe to Daliana's newsletter on www.dalianaliu.com for more on data science.

Jeremy's Twitter: https://twitter.com/JvNixon

Jeremy's Blog: https://jeremynixon.github.io/

Daliana's Twitter: https://twitter.com/DalianaLiu

Daliana's LinkedIn: https://www.linkedin.com/in/dalianaliu

Jeremy's LinkedIn: https://www.linkedin.com/in/jeremyvnixon

(00:00:00) Introduction 

(00:01:50) Research in Google Brain 

(00:03:37) How he got into Google Brain 

(00:07:56) His 3-month plan to learn ML 

(00:17:55) The 100-hour workweek 

(00:33:26) What if he is tired 

(00:39:59) Why he found Omni 

(00:44:24) Data science problems in Omni 

(00:54:42) Future of machine learning 

(00:57:51) Silicon Valley is very accessible 

(00:59:47) The golden handcuffs 

(01:06:58) From data scientist to full-stack engineer 

(01:09:06) Close-minded data scientists 

(01:24:10) Advice to ML learners 

(01:29:41) Something he wished that he did when he was younger 

(01:37:25) The future of his career 

(01:42:17) Connect with Jeremy

Sergey Feldman is the head of AI at Alongside, providing mental health support for students. He is also a Lead Applied Research Scientist at Allen Institute for AI, where he built an ML model that improved search relevancy for scientific literature. Sergey has a PhD in Electrical and Electronics Engineering from the University of Washington. Today we’ll talk about machine learning for search, his consulting project for the Gates Foundation, AI for mental health, and career lessons. Make sure you listen till the end. If you like the show, subscribe, leave a comment, and give us a 5-star review. Subscribe to Daliana's newsletter on www.dalianaliu.com/ for more on data science.

Daliana's Twitter: https://twitter.com/DalianaLiuDaliana's  

Daliana's LinkedIn: https://www.linkedin.com/in/dalianaliu/  

Sergey's LinkedIn: https://www.linkedin.com/in/sergey-feldman-6b45074b/ 

Data Cowboys: http://www.data-cowboys.com/

Sergey Feldman: You Should Probably Be Doing Nested Cross-Validation | PyData Miami 2019: https://www.youtube.com/watch?v=DuDtXtKNpZs

December 4th, 2018 - Breakfast with WACh with Dr. Sergey Feldman, PhD: https://www.youtube.com/watch?v=vA_czRcCpvQ

(00:00:00) Introduction 

(00:01:24) Machine learning skeptic 

(00:03:02) Tree-based models for search relevance 

(00:14:34) How to do error analysis 

(00:19:20) Nested cross-validation 

(00:21:34) Model evaluation 

(00:30:43) Error analysis common mistakes 

(00:33:37) How to avoid overfitting 

(00:35:56) Consulting project with Gates Foundation 

(00:41:16) Tree-based models vs linear models 

(00:45:19) Working with non-tech stakeholders 

(00:50:20) Chatbot for teen’s mental health 

(00:54:32) Can ChatGPT provide therapy?  

(00:58:12) How he got into machine learning 

(01:02:12) How to not have a boss 

(01:03:46) Feelings vs Facts 

(01:09:02) Future of machine learning 

(01:11:30) How to prepare for the future 

(01:13:39) AutoML 

(01:17:12) His passion for large language models

Philip Tannor is the Co-Founder and CEO of Deepchecks, a python package to run checks for machine learning models. Previously, he was the head of data science group at the Isreal Defense Force. He has a master's degree from Tel Aviv University in engineering, his thesis was about a new algorithm that combines neural networks with gradient-boosting decision trees. Today we’ll talk about his career journey, how to build your data science muscle memory, the algorithm he worked on, and how to check ML models. If you like the show subscribe to the channel and give us a 5-star review. Subscribe to Daliana's newsletter on www.dalianaliu.com/ for more on data science and career.

Daliana's Twitter: https://twitter.com/DalianaLiuDaliana's 

LinkedIn: https://www.linkedin.com/in/dalianaliu/ 

Philip’s LinkedIn: https://www.linkedin.com/in/philip-tannor-a6a910b7/?originalSubdomain=il

Augboost: https://medium.com/@ptannor/augboost-like-xgboost-but-with-few-twists-e4df4017a5c4

(00:00:00) Introduction 

(00:01:17) How did he get into ML 

(00:02:52) Data science in the military 

(00:08:15) How to take feedback 

(00:13:24) Handling criticism 

(00:15:12) What he worked on 

(00:18:18) testing deployment 

(00:21:28) How to build the data science muscle memory 

(00:27:09) Improving the skills of data scientists 

(00:30:42) His thesis in grad school 

(00:36:59) Combine NN and gradient boosting 

(00:40:05) Aug boost 

(00:41:15)Tools he uses 

(00:45:58) Deepchecks 

(00:50:46) Most challenging part of building Deepchecks 

(00:52:05) How can people contribute 

(00:53:40) Behind the scenes 

(00:56:09) Deciding how to fix or improve the model 

(01:00:49) Advise for those who wanna create open-source projects 

(01:04:07) Features to add for the enterprise product 

(01:06:57) About his life and career right now 

(01:08:27) Connect with Philip

Who is Daliana? This is a conversation I had in 2021 with Harpreet Sahota. I talked about my unexpected journey to data science all the way back in high school, things I wish I could know earlier about my career, the projects I worked on, what is like to be a quote-and-unquote influencer on Linkedin, and more. If you want more content from me, I write about data science and career nerdy jokes, on my Linkedin and you can subscribe to my very infrequent newsletter at dalianaliu.com. I’m curious what you think about this episode, leave a comment on YouTube or send a DM on Linkedin. Hope you enjoy the Daliana special!  

Daliana's Newsletter: https://dalianaliu.com 

Daliana's Twitter: https://twitter.com/DalianaLiu 

Daliana's LinkedIn: https://www.linkedin.com/in/dalianaliu/ 

Harpreet's LinkedIn: https://www.linkedin.com/in/harpreetsahota204/ 

The artist of the data science podcast: https://theartistsofdatascience.fireside.fm/ 

(00:00:00) Introduction 

(00:02:52) Where did Daliana grow up 

(00:05:19) Daliana in highschool 

(00:07:11) How did she got into data science 

(00:11:36) Why is writing important for data scientist 

(00:15:51) How to write better 

(00:20:56) Career lessons you didn't learn in school 

(00:27:40) Imposter syndrome 

(00:31:29) Day-to-day work as a data scientist 

(00:36:16) Most common mistakes data scientists make 

(00:39:41) Data Analyst vs. Data Scientist 

(00:42:30) What is the science in data science? 

(00:44:51) Can everyone be a data scientist 

(00:49:21) Linkedin profile tips for job search 

(00:52:59) How she creates content 

(00:54:11) Being a data scientist "influencer" 

(00:56:04) Why she started "the data scientist show" 

(01:01:16) Women in data science 

(01:06:39) What's her legacy 

(01:09:43) What is she reading 

(01:14:21) Connect with Daliana