In this episode of Leading with Data, we delve into the key insights shared by Vijay Karunamurthy, a seasoned AI professional with a rich background in engineering and machine learning. From his pioneering work at YouTube and Apple to his current role as Field Chief Technology Officer at Scale AI, Vijay offers invaluable perspectives on the evolving landscape of artificial intelligence. Join us as we explore how AI transforms various sectors and the essential skills needed for success in this dynamic field.

You can listen to this episode of Leading with Data on popular platforms like Spotify, Google Podcasts, and Apple. Pick your favorite to enjoy the insightful content!

Key Insights from our Conversation with Vijay Karunamurthy

• The integration of AI in various sectors, such as finance, is leading to significant efficiency gains by assisting professionals with complex tasks like SQL query optimization.
• Real-world testing and evaluation of AI models are essential for understanding their practical applications and limitations, guiding responsible AI deployment.
• AI agents represent a significant trend, with multiple models working collaboratively to tackle complex problems more effectively.
• The AI landscape is rapidly evolving, with new research and capabilities emerging regularly, making it crucial for professionals to stay informed and adaptable.
• Aspiring AI professionals should focus on foundational skills and practical applications of AI, rather than trying to keep up with the overwhelming pace of research.

Join our upcoming Leading with Data sessions for insightful discussions with AI and Data Science leaders!

Now, let’s look at the details of our conversation with Vijay Karunamurthy!

How did your journey in AI begin?

My journey into AI began during my undergraduate years when AI was just starting to gain traction. As a biochemistry major, I was fascinated by the biological processes of living organisms and the idea of replicating these processes through neural networks. My first project involved manually coding a neural network from a book, which was both challenging and thrilling due to the lack of resources like GitHub at the time. This initial foray into machine learning, despite the subsequent “neural network winter,” laid the foundation for my career in AI.

What was it like working on AI at YouTube and Apple?

At YouTube, the challenge was to connect viewers with relevant content amidst the massive daily uploads. Initially, we relied on user-provided data like video titles and comments, but eventually, we leveraged convolutional neural networks to understand video content directly. This was a game-changer, allowing us to recommend videos with little to no user input. At Apple, the focus shifted to privacy-preserving machine learning, which was a different but equally fascinating challenge. We explored how machine learning could enhance content discovery across various media on users’ devices while respecting data privacy.

Can you share an example of AI’s impact in the financial sector?

Certainly! We collaborated with a financial services company to assist their data analysts. By fine-tuning a language model, we enabled it to suggest more efficient SQL queries, leveraging indices and optimizing table joins. This AI assistance reduced the time analysts spent rewriting queries from hours to mere minutes, significantly boosting productivity.

How do you approach testing and evaluation of AI models?

Testing and evaluation are crucial. We focus on real-world scenarios rather than synthetic benchmarks to understand how models perform in practical applications. This involves creating diverse data distributions that reflect actual use cases and continuously refining our evaluation benchmarks. By doing so, we can discern the trade-offs between large and small models and make informed decisions about which is best suited for a given scenario.

What are the current trends in AI that excite you?

The rise of AI agents, where multiple models collaborate to solve problems, is particularly exciting. This architecture allows for sophisticated, parallelized problem-solving across various data types. Another trend is the evolution of coding AI, where models are becoming adept at assisting developers with complex coding tasks, significantly reducing the time and effort required to make code usable.

What advice would you give to aspiring AI professionals?

Focus on the fundamentals of AI, such as testing, evaluation, and pipeline construction. Stay current with research but don’t feel pressured to keep up with every paper—select a few each week that are most relevant to your interests. Remember, the field is rapidly evolving, so continuous learning and adaptability are key.

Summing-up

 In our conversation with Vijay Karunamurthy, we’ve gained crucial insights into the ever-changing world of artificial intelligence. From the profound impact of AI in sectors like finance to the importance of real-world testing and evaluation, Vijay’s experiences shed light on the challenges and opportunities in this rapidly evolving field. Aspiring AI professionals can draw inspiration from Vijay’s journey and his advice to focus on fundamentals, stay adaptable, and continuously learn in order to navigate the complexities of AI and lead with data-driven solutions.

For more engaging sessions on AI, data science, and GenAI, stay tuned with us on Leading with Data.

Check our upcoming sessions here.