2017 - Now
Applied Research Scientist at Amazon
Here's where things get interesting. I get to work on ML problems that span everything from understanding human language to analyzing how people interact with systems. It's like being a detective, but instead of solving crimes, I'm solving puzzles about how to make machines understand and help humans better.
Currently, I'm leading efforts to revolutionize how we support sellers using ML—because even in the age of AI, human problems still need human-centered solutions.
Here's where things get interesting. I get to work on ML problems that span everything from understanding human language to analyzing how people interact with systems. It's like being a detective, but instead of solving crimes, I'm solving puzzles about how to make machines understand and help humans better.
Currently, I'm leading efforts to revolutionize how we support sellers using ML—because even in the age of AI, human problems still need human-centered solutions.
2016
ML Intern at Akamai - Ghost Security Team
Yes, the team was actually called "Ghost Security"—how cool is that? I helped them turbocharge their Web Application Firewall using ML. We went from 1ms to 20 microseconds per request. That's a 50x speed-up!
Turns out, teaching machines to spot bad actors is surprisingly similar to playing whack-a-mole, except the moles are constantly evolving.
Yes, the team was actually called "Ghost Security"—how cool is that? I helped them turbocharge their Web Application Firewall using ML. We went from 1ms to 20 microseconds per request. That's a 50x speed-up!
Turns out, teaching machines to spot bad actors is surprisingly similar to playing whack-a-mole, except the moles are constantly evolving.
2015 - 2017
Graduate Research at UC San Diego
This is where I fell in love with making machines "see" and "understand." Spent my days (and many nights) teaching robots how to navigate the world, working on everything from self-driving cars to health robotics.
My favorite project? Figuring out how to detect motion when the camera itself is zooming around on a drone. It's like trying to spot a moving car while you're on a roller coaster—except with math and algorithms.
This is where I fell in love with making machines "see" and "understand." Spent my days (and many nights) teaching robots how to navigate the world, working on everything from self-driving cars to health robotics.
My favorite project? Figuring out how to detect motion when the camera itself is zooming around on a drone. It's like trying to spot a moving car while you're on a roller coaster—except with math and algorithms.
2014 - 2015
Software Engineer at Samsung R&D
Before diving into ML, I was building features for millions of Samsung users. Created some neat calendar features that probably helped someone remember their anniversary (you're welcome!).
The highlight? Working on folding phone prototypes at Samsung HQ in Korea. This was before they were cool— I like to think I was part of making the future unfold (pun intended).
Before diving into ML, I was building features for millions of Samsung users. Created some neat calendar features that probably helped someone remember their anniversary (you're welcome!).
The highlight? Working on folding phone prototypes at Samsung HQ in Korea. This was before they were cool— I like to think I was part of making the future unfold (pun intended).
2010 - 2014
B.Tech at NIT Surat
Where it all began! Got my hands dirty with computer vision research, literally building 3D worlds from Kinect sensors and RGB cameras. This was back when Kinect was the coolest tech around (remember those?).
Learned CUDA programming to make everything run faster—because waiting for your code to run is the worst kind of suspense. Also discovered that debugging GPU code is like debugging regular code, but in parallel universes simultaneously.
Where it all began! Got my hands dirty with computer vision research, literally building 3D worlds from Kinect sensors and RGB cameras. This was back when Kinect was the coolest tech around (remember those?).
Learned CUDA programming to make everything run faster—because waiting for your code to run is the worst kind of suspense. Also discovered that debugging GPU code is like debugging regular code, but in parallel universes simultaneously.