Create a data and ML product that users love - DSBoost #7
Discover more from DSBoost
Create a data and ML product that users love - DSBoost #7
Welcome to the seventh issue of DSBoost, the weekly newsletter where you can discover exciting 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 Hector Haffenden, who is a Lead Data Scientist in management consulting. Enjoy:
What did you study/are you studying?
I did an undergraduate in financial mathematics.
What are your favorite resource sites and books (ML/AI)?
For me, this is the data science starter kit:
Hands on Machine Learning by Aurélien Géron
Udemy for tactical upskilling
Azure, GCP and AWS documentation for cloud
What got you into your current role (portfolio, certification, etc.)?
I joined a management consultancy out of university, through my interest and projects in data science, I worked with our leadership teams to create a new data science specialist pathway - which is how I ended up in my current role.
What do you enjoy the most in your work?
Working hand in hand with a variety of clients, and finding - then solving their biggest problems.
Often data science projects don’t focus on the most critical challenges companies are facing. I enjoy looking at the data and understanding clients’ challenges before jumping into solutions - this leads to a bigger impact and a more bought in client.
What tools do you use the most / favorite tools?
VSCode, Python, Azure is where I spend most of my DS delivery time.
VSCode is hard to beat with its Microsoft integrations and flexibility to different programming languages. Having the right tech setup is helpful in moving towards developing production code.
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?
I’ve been using GitHub Copilot since it was released, and I'm a big fan. It probably completes ~50% of my code.
ChatGPT has replaced a significant amount of my Stack Overflow use, and anticipate this to continue. There are edge cases (recently found quite a few when developing in Azure) that ChatGPT gets wrong, but overall it saves time.
What is your favorite topic within the field?
Reinforcement Learning and recommendation engines:
RL is interesting to me as there not a lot of knowledge about the field in industry, and there are a lot of opportunities to optimise processes.
Recommendation engines are great to learn as they ensure you have a solid fundamental understanding of data science / ML - and cut across NLP, tabular and Image ML.
Which one of the recent AI/ML models will have the most significant impact on the industry in your opinion?
We are starting to see a push from leaders in business to start integrating Large Language Models into products. However, that is still early. From what I’ve seen it is still GBMs that allow new data scientists to develop high performing models - which allow your tech leads to focus on evaluation, bias and integration of your models with the business.
As an aside, I’ve seen the creation of multi-purpose data assets skyrocket companies ability to have a significant impact with AI.
What are you currently learning or improving (topics you are interested in nowadays)?
Cloud, AI ethics and integrating LLMs in production.
What is the biggest mistake you've made? (preferably DS related)
Key themes are: Choosing a metric that doesn't align with client KPIs, assuming data quality, assuming all data scientists use solid evaluation methods
What is your most significant achievement? (preferably DS related)
Since I started working in data science, I’ve done one hackathon a year (to balance other projects, work, personal life), this has been vital for my development and I’m happy with my achievements in those.
Can you share a fun fact about yourself?
I initially applied to university to do Psychology, I think it’s one of the most interesting fields, and I like to stay up to date with it.
There is a lot of opportunity in the application of data science to psychology which is exciting to follow and engage with.
🎙️ Podcast of the week
SuperDataScience 658: How to Build Data and ML Products Users Love
Key takeaways:
To build successful data products, focus on user and business outcomes, and develop machine learning power products that users love.
Many data projects fail due to poor planning, lack of communication, and unrealistic expectations.
Successful data product development requires a diverse team with a range of skills, including data science, engineering, design, and product management.
To succeed in data product management, cultivate strong communication and collaboration skills, stay up-to-date with emerging technologies and industry trends, and maintain a user-centric mindset.
Thanks for reading DSBoost! Subscribe for free to receive new posts and support my work.
🧵 Featured threads
🤖 What happened this week?
Microsoft Germany CTO Andreas Braun mentioned at an AI kickoff event on the 9th of March 2023 that OpenAI’s GPT-4 was going to be released this week.
Wonder Studio is a new AI tool that automatically animates, lights and composes CG characters into a live-action scene; all you need is a camera!
Midjourney plans the release of version 5 soon. The first alpha images show very promising results, which means that they are ready to take generative AI one step further.
👥 Under the radar
Words from Gibson Hurst:
Hey, I'm Gibson Hurst. I'm currently pursuing a degree in Statistical Science and Data Analytics. I'm excited to apply my knowledge and skillset as an incoming Marketing Analytics Intern at a large financial company. Outside of class and work, I share data analytics related content on Twitter to help others entering the field like me. The online community around data analytics has been so supportive of my journey and I'm thankful for the opportunity to give back by helping others in this space.