3 routes Data Scientists can choose from - DSBoost #10
Discover more from DSBoost
3 routes Data Scientists can choose from - DSBoost #10
Welcome to the 10th 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!
💬 Interview of the week
This week we interviewed Sasi, who is a Data Analytics Consultant. Enjoy:
What did you study/are you studying (if your background is different from DS, how did you end up in the field)?
I did my undergrad in Mechanical engineering. Then I did my postgrad in Marketing and Analytics. During my postgrad I learned a lot of Data Science and Analytics knowledge and tools which help me now.
What are your favorite resource sites and books (ML/AI)?
What got you into your current role (portfolio, certification, etc.)?
Several portfolio projects in the area of market research, customer satisfaction analysis etc.
Having done my diploma in Marketing and Analytics also helped.
What do you enjoy the most in your work?
Advanced and complex chart and graph requests from clients in dashboards are always challenging and I like developing them.
What tools do you use the most / favorite tools?
Used the most in recent times is TIBCO Spotfire Analyst and most favorite has to be SQL.
Do you use ChatGPT or other Al tools during your work? If so, how do they help you? Do they change your approach to problems?
ChatGPT is great in explaining complex code which was written by someone else. Makes my entire process really efficient. Also when there is an error or when I have to use multiple functions together, I just describe the task using prompts to ChatGPT and it helps me out. It a great tool to make you more efficient in general.
What is your favorite topic within the field?
I really like the ML application around Marketing and customer behavior. The insights which you can draw from the analysis always help to reveal something which is really interesting!
Which one of the recent AI/ML models will have the most significant impact on the industry in your opinion?
I think GPT4 will have a major impact on how we go forward in the area of tech in general and as well as marketing. A lot of new possibilities.
Several smaller firms started to adopt to cloud computing (AWS, Azure). So might be a good thing to pick up if you want to have a career in Data Analytics.
What are you currently learning or improving (topics you are interested in nowadays)?
In general, I am trying to add more interesting projects in my project portfolio.
What is the biggest mistake you've made? (preferably DS related)
Once I was cleaning the database and dropped a view which was the backend for a dashboard without taking a backup. This set me back in my deliverables by 1 week.
What is your most significant achievement? (preferably DS related)
I was given a “Kudos” Award within my firm for learning the new tools quickly for a particular client engagement and meeting the strict deadline.
Can you share a fun fact about yourself?
In my undergrad I didn't like coding! My perspective about it has changed drastically over the past 3-4 years and now I don't go a single day without writing a piece of code/script!
🎙️ Podcast of the week
SDS 665: How to be both socially impactful and financially successful in your data career by Jon Krohn. Guest: Josh Wills
Key takeaways:
A data scientist is a person who is better at statistics than software engineers and better at software engineering than statisticians.
Data scientists go in one of three different directions:
One direction is going deep into the methodology, like AB testing, causal inference, and contextual bandits. They have these hard statistical problems.
Another route is to become a very deep domain expert in a problem area. For example, becoming an expert in SaaS pricing models. How does it translate into core business metrics of MRR, and ARR, and how does it relate to our enterprise deals? They are in the middle.
The final path is the data engineer, with a little bit of statistics, but mainly software engineering.
Josh: I learned at Google that simple analysis on top of high-quality data usually wins the day. Working with extremely high-quality data and utilizing straightforward methods is much more convenient than dealing with subpar or partially relevant data and applying complex techniques to enhance or optimize it.
Catastrophic machine learning failures are typically data quality problems.
🧵 Featured threads
🤖 What happened this week?
LumaLabs has brought NeRF technology to Unreal Engine, allowing users to capture realistic scenes and use them as environments in their projects.
Midjourney has introduced a new command, /describe, which provides users with four suggestions for text that best describes a specific image.
Stable Diffusion XL (SDXL), a 2.3 billion parameter variant of Stable Diffusion 2.1, is being tested and refined before its open-source release. It will feed into Stable Diffusion 3.
BloombergGPT, an LLM for finance, trained on the largest domain-specific dataset with 363 billion tokens augmented with 345 billion tokens from general-purpose datasets. It outperforms existing models on financial tasks without sacrificing general LLM benchmarks.
The fight against AI begins: First, tech leaders call for a pause of GPT-4.5 and GPT-5 development due to ‘large-scale risks’. And then, ChatGPT to be temporarily banned in Italy due to privacy concerns.