“My fascination with AI began when I first heard about IBM’s supercomputer, Deep Blue, defeating Garry Kasparov.”
Analytics India Magazine (AIM) got in touch with Hamsa Buvaraghan for this week’s ML practitioner series. Hamsa currently leads the Google Cloud data science and MLOps solutions team, developing revolutionary software solutions for business problems with Google’s data analytics and AI / ML products. She holds a degree in computer science from Mysore University and an MBA, Honors from Saint Mary’s College of California.
AIM: How did your fascination with AI begin?
Hamsa: My fascination with AI began when I was in India in 1997 when I heard about IBM’s supercomputer Deep Blue, which defeated Garry Kasparov. That made the headlines back then. After that, I wanted to know more about it. However, access to research was very difficult at the time because I had neither a computer nor access to the Internet. I was introduced to computers by my father when I got access to a computer in his office at the age of 10. The first thing I researched back then was Lotus Notes. At the suggestion of my parents, I later studied computer science.
Later, when I started working, I read several IEEE research papers. I read newspapers like smart games; Beyond the Deep Blue horizon, Deep Blue’s hardware-software synergy. At the time, I was not only fascinated by AI, but also by the application of AI to solving real problems. I was also excited about Biomedical Engineering, which led me to books on Neural Networks and AI for Biomedical Engineering and articles on Training Neural Networks for Computer-Aided Diagnosis. When it comes to machine learning, I’m largely self-taught.
AIM: What were the initial challenges?
Hamsa: At that time I only knew the C programming language from university. The book “C Programming Language” by Brian W. Kernighan, Dennis M. Ritchie, was my valued resource. I later taught myself C ++, Java, C # and Python. The greatest challenge was access to resources. The other challenge was the lack of easy-to-use ML frameworks like TensorFlow. I started with WEKA, which was an easy decision back then. I also used Deeplearning4j and MALLET. At that time, my team and I had to implement many algorithms ourselves. Today, developer life is so much easier with TensorFlow, Keras, and other tools.
GOAL: Tell us about your role here. What does a typical day look like?
Hamsa: I lead the AI solutions team at Google Cloud. I drive vision, mission and strategy in three main solution areas – Data Science, MLOps and Large Scale ML. Working with cross-functional teams, I help define AI solutions and their growth strategy for this business at Google Cloud. Our team solves the complex / big AI / ML problems for our clients.
Every day is very different. On a typical day, I could meet with corporate clients and their data science teams, or define the executive’s vision and direction for AI solutions, lead and lead the team in prioritization, coordinate with engineering directors, product managers, sales strategy and senior executives , Brainstorming / design thinking to encourage thinking outside the box and develop the 10X ideas, and evangelism at conferences for clients and partners. The most important thing for me is that every day on Google is like opening a new page to learn something. That’s the great thing about working here.
AIM: How do you approach a data science problem?
Hamsa: I’m interested in the application of data science and AI to real problems. However, I strongly recommend not looking at every problem through the lens of data science. I step back and begin to understand the problem. Then I validate whether it’s a data science problem. If so, I will clearly articulate this as a data science problem. I also check if the solution to a particular problem can be generalized to apply to other similar problems and in other related scenarios. For example, our team at Google Cloud recently developed a solution for a customer use case for real-time matching and retrieval, and this included designing, implementing and automating article-to-article recommendation systems using Google Cloud products like BigQuery ML, Dataflow, Data storage and Vertex AI. This system can also be used to address a wide range of customer use cases for real-time matching and retrieval. For example, finding relevant products, replacing certain products – think of online grocery shopping, finding songs, websites, etc.
GOAL: What does your machine learning toolkit look like?
Hamsa: I prefer frameworks like TensorFlow, Keras. After doing all the hard work of building things from scratch, today I tend to use more modern models. At Google Cloud, I tend to use BQML, Vertex AI for training and prediction, and Kubeflow pipelines for MLOps.
AIM: Which AI domain do you think will prevail this decade?
Hamsa: I recently read an article from Gartner about the “Hype Cycle for Artificial Intelligence 2020″. Despite Covid-19, 47% of AI investments have remained unchanged and companies are instead realizing how AI can create more value. This wouldn’t have been possible if AI / ML was just a hype. I see very strong potential in AI / ML in the 10 years. I also see tech companies / creators moving towards the democratization of AI / ML. Thus use cases, applications that are more easily used by broader users, will prove themselves over time. This is very important for faster adoption of AI in industries that are just starting out with AI / ML. We need to hide the complexity of AI from the end users. For the next 10 years, almost every industry will be pouring AI into everything they do, not because of the hype but because of the real value they see with AI.
AIM: What is your advice to aspirants looking to land ML roles on Google?
Hamsa: If you look at the roles of data scientists, typically you need to enable insights and data-driven decisions, design and support experiments, and translate analytical results into actionable product or business recommendations. You will work with a multidisciplinary team of engineers, analysts and product managers on a wide variety of problems. And remember, when it comes to Google, it’s always to scale. We have so many products with over a billion users. You need to bring analytical rigor to your approach. Practice by playing around with developing, organizing, and performing experiments on large complex data sets, solving difficult non-routine analysis problems, and using analytical methods. Do an end-to-end analysis. For example, be part of competitive platforms like Kaggle, which are a great introduction to these challenges. There is a lot of complex content. Don’t be afraid to be a part of it, so you can learn and develop your skills.
Last but not least, network. Attend conferences where you can meet like-minded people, find a mentor and, most importantly, pursue your passion.
My top book recommendations for beginners:
- Deep Learning (Adaptive Computation and Machine Learning series) by Ian Goodfellow, Yoshua Bengio and Aaron Courville.
- Machine Learning Longing by Andrew Ng
- Hands-on machine learning with Scikit-Learn and TensorFlow by Aurélien Géron
- Data Science on the Google Cloud Platform Lak Lakshmana
I also recommend all Andrew Ng Coursera courses related to AI / ML, especially Machine Learning Engineering for Production (MLOps) Specialization.
Join our telegram group. Become part of a dedicated online community. Join here.
Subscribe to our newsletter
Get the latest updates and relevant offers by sharing your email.