10 Takeaways about the Job Market, LLMs & Future of Data Science- DSBoost #22

David Andrés

and

Levi

Jun 27, 2023

Welcome to the 22nd 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

Q&A: Job Market, LLMs, Future of Data Science - KNN Ep. 155

Key takeaways:

The field of data science is evolving, with a trend of specialization within the domain.
Data science work is being broken down into specific components like data engineering and machine learning engineering, leading to fewer generalist data science roles.
Hands-on experience, such as real-world projects, is valuable for landing a data science job, especially for entry-level positions.
Transitioning into data science from a different field is feasible but requires significant effort, and exploring subcategories like data analytics or software engineering can provide entry-level opportunities.
Maintaining consistency and motivation while learning data science is a common challenge, and pursuing personally interesting and challenging projects can enhance learning and skills.
Large language models (LLMs) like ChatGPT can be a valuable tool for coding tasks in data science.
Learning how to effectively use tools like LLMs can enhance the capabilities of data scientists and provide a significant return.
Burnout is a common experience in data science work, and taking breaks and engaging in other activities is important to avoid becoming overwhelmed.
Keeping up with technical skills in data science can be achieved through project-based learning, educational content, and learning from other practitioners in the field.
The roles of data scientists and data engineers are still relevant and important, despite the assistance provided by tools like LLMs.

💬 Interview of the week

This week we interviewed Rohan, who is a Machine Learning Engineer & Kaggle Master. Enjoy:

What did you study/are you studying (if your background is different from DS, how did you end up in the field)?

By academics I am MBA-Finance with International Experince in I-Banking. But I left banking to learn and work Software Engineering.

What are your favorite resource sites and books (ML/AI)?

Kaggle and Twitter.

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

Portofolio work (Github / Kaggle / Linkedin Posts / Twitter Posts ). My YouTube channel also definitely helped.

What do you enjoy the most in your work?

When model.predict() magically outputs a very reasonable answer.

What tools do you use the most / favorite tools?

PyTorch and XGBoost Library

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?

Yes GPT4 can explain some script very well. With beautiful examples. Also some boilerplate code like Unit Tests are created super fast by GPT4.

What is your favorite topic within the field?

Currently the magical power of VectorDB + Langchain.

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

For foreseeable future Transformer and QLoRA.

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

RAG (Retrieval Augmented Generation)

What is the biggest mistake you've made? (preferably DS related)

Ignoring some Deployment related skill development.

What is your most significant achievement? (preferably DS related)

When my ML YouTube Channel reached 5k Subscriber after posting almost 250 videos.

Can you share a fun fact about yourself?

I can't resist playing around with AI-based face-swapping apps. I once accidentally sent a company-wide email with my face swapped onto our CEO's, and it became an office meme.

🧵 Featured threads

🤖 What happened this week?

DeepMind, the AI lab owned by Google, is developing a new AI system called Gemini, which aims to surpass the capabilities of OpenAI's ChatGPT. Gemini combines the strengths of DeepMind's AlphaGo program with the language capabilities of large models like GPT-4. The system, still in development, will incorporate techniques such as reinforcement learning and tree search to enhance its problem-solving and planning abilities. Gemini is part of Google's response to the competitive threat posed by generative AI technologies. DeepMind's CEO, Demis Hassabis, acknowledges the need for research on the risks and controllability of more advanced AI models and expresses a desire to make their systems more accessible to outside scientists.

AI and deep learning didn't cause the job apocalypse many feared! New data reveals that industries benefiting from AI actually saw a 5% increase in skilled workers. Turns out, technology creates more demand for expertise while taking over mundane tasks. However, the impact of generative AI, like ChatGPT, still remains uncertain. While some jobs, like copywriters, have been hit, AI systems are still flawed. So, don't fret about your career just yet—past predictions of AI wiping out jobs haven't quite come true.

MosaicML just released their most advanced language model yet, the MPT-30B series. They're like turbocharged versions of GPT-3, offering better quality and affordability. Businesses can harness the power of AI while keeping data secure. Replit, Scatter Lab, and Navan are already using them for code generation, chatbots, and travel management. Whether you're a developer or need a hassle-free solution, MPT-30B is a game-changer in language models, giving businesses the AI edge they need.