How to be a Data Scientist That People Trust - DSBoost #2

Welcome to the second issue of DSBoost, the weekly newsletter where you can discover interesting people in the ML/AI world, get the main takeaways of a relevant podcast, and stay up to date with the latest news in the field!

🎙️ Podcast of the week

SDS 645: Machine Learning for Video Games - Podcast Guest: Carly Taylor

Key takeaways:

• Anomalies can be very important, and it's essential to be able to identify them and figure out why they're happening.

• The most desired skill in Data Science is effective communication! That helps every data scientist become more effective and be seen as a more trusted, collaborative partner, as opposed to the person who knows some magic in Python.

What gives you the ultimate power is making other people feel like they can understand what you do and empowering them to think like, hey, I can actually understand Data Science.

💬 Interview of the week

This week we interviewed Dan, who is a Machine Learning Engineer at Infineon. Enjoy:

What did you study?

I have a Master's degree in Electronics and Telecommunications. I started off my career as a software test engineer. Software testing was manual and tedious, so I often used my programming skills to automate and optimize my work.

Later I was given an opportunity to work on a machine learning project that would improve how testers analyze logs. This jumpstarted my career as a machine learning engineer.

What are your favourite resources?

• Designing Machine Learning Systems by Chip Huyen.

• Hands-on Machine Learning with Scikit-Learn, Keras, and TensorFlow by Aurélien Géron.

•  I also monitor Kaggle for any fun projects to work on.

What got you into your current role (portfolio, certification, etc.)?

Mainly the experience from the ML projects that I have worked on. I was responsible for the entire machine learning lifecycle. With a wide range of skills, you can qualify for most ML roles.

What do you enjoy the most in your work?

ML models generating predictions based on a set of data still amazes me. Creating a system that utilizes this amazing technology to solve a problem that has a major impact on a business is really a rewarding experience.

Do you use ChatGPT or other Al tools during your work?

Sometimes I might ask ChatGPT when I have some code errors, but it doesn't really give reliable answers. Stackoverflow is still better.

What is your favorite topic within the field?

Natural Language Processing, because I have worked a lot with log analysis & classifying developer tickets.

Which one of the recent AI/ML models will have the most significant impact on the industry in your opinion?

I think AlfaTensor by DeepMind is going to have a huge impact. With it we can dramatically decrease the training time of an AI model.

What are you currently learning or improving (topics you are interested in nowadays)?

Currently, I am trying to improve my knowledge of machine learning system design. Building ML models is only the tip of the iceberg. Without having the right infrastructure and systems in place to streamline the entire ML building process, it is very hard to work as an ML Engineer or Data Scientist.

What is the biggest mistake you've made?

At the beginning of my career, the biggest mistake I made was being too confident in the results of my ML model and reporting its success too quickly to management.

What is your most significant achievement?

My biggest achievement was being able to successfully deploy a machine learning model into production that provided real results for my team. The model was able to find some bugs that the testers did not find.

Can you share a fun fact about yourself?

I am a huge fan of martial arts. I use to train in Hapkido, a Korean martial art.

🧵 Featured threads

Thomas Sorheim @ThomasSorheim

Major leak!! 🚨 Microsoft integrates ChatGPT in Bing. This looks crazy! See all the SCREENSHOTS in this thread 🧵👇 #chatgpt #ai #bing #microsoft #leak

Eric Wallace @Eric_Wallace_

Models such as Stable Diffusion are trained on copyrighted, trademarked, private, and sensitive images. Yet, our new paper shows that diffusion models memorize images from their training data and emit them at generation time. Paper: arxiv.org/abs/2301.13188 👇[1/9]

🤖 What happened this week?

• OpenAI releases ChatGPT Plus. The monthly subscription plan for $20 offers benefits such as unrestricted access to ChatGPT, faster response times, and priority access to new features and upgrades.

• Runway announced Gen-1, a new Al model that is capable of Video-to-Video generation by using language and images.

👥 Under the radar

Here are a few words from Jean personally:

Hi there! I'm Jean Stevany, a lecturer in Accounting by day and a Microsoft Student Learn Ambassador by night (or whenever I get a chance to dive into the world of data and technology). My passion for data is contagious and I love to share insights and knowledge about Power BI. As an educator, I'm on a mission to build a community of individuals interested in learning about data and technology. Join me as I explore the exciting world of Power BI and discover new ways to analyze and understand data. Let's learn and grow together!