Welcome Dr. Jaya Prakash (JP) Champati! (an Interview with Dr. Yun Lu)

I had a wonderful interview with Dr. Jaya Prakash (JP) Champati, one of our newest faculty members. In this interview, I learn about his experience as a research faculty in Madrid during the pandemic, and delve into his research in edge computing. 

 

 

Can you give me a quick introduction of your academic journey? 

JP: I am from the Andhra Pradesh state in India and I grew up in a coastal village. I went for undergrad in National Institute of Technology in Warangal, and masters at Indian Institute of Technology Bombay. I then worked in Broadcom Communications in Bangalore, in a different state in South India. I did my PhD at U of Toronto, with my supervisor Ben Liang, for five years. 

After five years at U of T, I went to Stockholm for a postdoc. This is at the KTH Royal Institute of Technology. I spent three years there, then I got a couple of offers ---one at University of Glasgow and another at the IMDEA Networks Institute in Madrid. 

Yun: So how did you choose between Glasgow and Madrid? 

JP: University of Glasgow is a well-established university, but IMDEA Networks has publications in the conferences I am publishing in, like INFOCOM. The group in Madrid was strong as well, and the good salary and the weather helped too. 

 

What was your experience when you first started your position in IMDEA, Madrid? 

JP: I started there roughly in November or December 2020.  

Yun: Wow, right smack in the middle of COVID. 

JP: Actually, it’s interesting---I went there for an interview on March 4th , then on March 8th everything shut down in Spain. When I was there things were still operating.  Everyone had started wearing masks, but with the hope that this can be contained. No one knew Spain would shut down on March 8th.  

When I started, I couldn’t meet many people, but I could still work. But I cannot complain because it was universal, affecting everyone. Still, everyone lost their social life.  

Yun: So how did you manage your first few months there? 

JP: My difficulties were not because of my work, but when I started in Madrid, my father had an accident. He had a brain hemorrhage, and he’s still affected even now. At that time, I stayed with him in the hospital for 1 1/2 months. And by the time I came back, there was nothing much else to think about, except for work, and just talking every day with family. My mind was occupied. 

After I came back to Spain, it's been almost one year of COVID. Spain went through three waves of it. By that time, there was a kind of norm. People are showing up. We were not following the rules as much. My colleagues were great, and we started going out for lunch, starting in summer 2021.  

How did you get into computer science? 

Until I entered into PhD, I worked in electronics and communications. My supervisor gave me the topic of edge computing. I was looking into scheduling jobs between mobile device and a computer or a server in a nearby spot. This was the main vision for Mobile Edge computing---to have computing resources at base stations so mobile devices can offload  computing jobs. 

Yun: What kinds of computing jobs? 

JP: For example, in mobile gaming, some of the games are better offloaded to a server. But this idea never flew. Network operators have a lot of inertia and won’t easily install these servers, and even if they do, developers must make the applications distributed. Even if the tasks are in a directed acyclic graph, it’s a big challenge to efficiently run it, because there are dependencies that make this a hard combinatorial problem. 

But for me, all this was mainly a scheduling problem. This is an area with roots in operations research and theoretical computer science---including the study of NP completeness. For example if a problem is really hard you first check whether it is NP complete, If it is, you have an excuse not to solve it!  

Now, I analyse online learning algorithms, prediction with expert advice and multi arm bandits. I would say that my background was not computer science, though I love algorithms, and the data structures and algorithms using C was my favourite class.  

What other activities did you enjoy during university? 

JP: I’m a good dancer! I don’t know traditional Indian dance, but I know the most popular ones, which is a mix between Western and folk dancing, like the ones you see in Bollywood movies. That’s where I learned it, though I didn’t have any professional teachers. I was actually the lead dancer in a group! We performed in cultural festivals, along with a small skit. 

 

Can you tell me about your research area? 

JP: My research area now is on getting big deep learning models to run efficiently on devices. In the current landscape there are so many deep learning models that are open source. Literally, hundreds of thousands of LLM's are available on HuggingFace. These can be deployed on phones, but they are small models, so the main aim is to improve the accuracy and efficiency---will it crash in two minutes? Is the response reliable? How powerful does my phone have to be? 

So the research I am focused on goes back to edge computing, where we get help from a server. Whenever you try to process queries on the phone, and there is not enough confidence in the response, we send it to the server. We take advantage of the bigger computer power, while not killing our smaller device’s batteries. 

Yun: Do people worry that the server will give back incorrect information—or worse, steal your data? 

JP: Of course! There are two potential solutions for this. One is if the server is dedicated, i.e., it’s my own machine, so it’s safe. Unfortunately, this is not always possible. Another way is to obscure the details of your query, to hide sensitive information. For example, instead of asking the model “Should my 7 year old son...” maybe ask “Should a 7 year old...”. There’s no perfect solution unfortunately, because you lose some information. 

Yun: Definitely, there’s always a tradeoff between data privacy and accuracy. 

 

What is the biggest challenge for you in your academic career. And how did you overcome it? 

JP: Definitely, it was my first three years of PhD and my first three years as an assistant professor. I submitted to a lot of top conferences and faced a lot of rejections. At the time I felt I failed my students, because I couldn’t give them the confidence they needed from an accepted paper.  

Yun: Rejection is such a universal experience as an academic. How did you keep motivation despite this setback? 

JP: I think I’m motivated for several reasons. The first reason is of course, I like research and I love analysing algorithms. The second reason is that I see the research funding I get as the taxpayer's money, so I feel the responsibility to excel.  

 

What advice would you give students facing uncertainty in their (academic) career? 

JP: I think my takeaway from my experiences is that you have to focus on small goals. And at some point, you will know what you want with your life and eventually have a long term goal and vision. As long as you keep doing your best, even if something doesn’t go as planned, it’s totally fine. 

 

Closing Question: If you could go back in time and give yourself any advice, what would it be?  

JP: I used to get scared of a lot of things. I didn’t necessarily doubt myself, but I do get anxious, and I tend to avoid conflicts. I would tell myself to stand up and question things when something doesn’t feel right.  

Yun: Have you encountered a situation where you wish you had said something, but didn’t? 

JP: Definitely, as a personal example, there was one time where an elder in my village said, very sarcastically, to not listen to my father, and was giving reasons why not. I didn’t understand at the time, but later I realised that he had meant to insult my father. He was laughing and joking, and everyone kept silent, so I didn’t say anything.  

Professionally, sometimes in faculty meetings, people would argue issues, and I feel that I should have expressed my thoughts instead of keeping silent!