As a graduate student at Purdue, I work at Predictive Science Lab under the guidance of Prof. Ilias Bilionis. My research revolves around utilizing machine learning techniques within probablistic frameworks to model physical systems. I am particularly interested in developing surrogate models and consolidating data, heuristics and physical constraints in a Bayesian manner. I usually write in python using PyTorch or JAX for developing the models, and run the programs on Purdue's superclusters such as Bell.
My thesis involves developing surrogate models for fields using neural networks and adopting theoretical ideas from Lie groups & algebra. I aim to use these findings to make meaningful contributions to Lilly's research effort at Purdue in the domain of intrathecal and injectable drug-delivery. I have previously worked on and applied methods such as Gaussian Process Regressions, Poisson Regression and Data-compression using libraries such as numpyro and GpyTorch.
My plan of study during my graduate school revolves around courses of statistics and machine learning. As a part of the machine learning course I wrote a Conditional-GAN model and a custom FID protocol to quantify the sensitivity of the model against hyper-parameters. The code can be found here. Courses, at glance:
I worked for 3 years at Bosch's research center in India: Technical Center India. Here as a trainee I worked on coordinating software requirements for digitization in the supply chain domain. Then I moved to working on electric vehicles and their firmware validation, before finally transitioning to my longest role as a research engineer developing automotive connectivity software.
During my time in the connectivity research team, our small team of inter-disciplinary engineers tasked with the end-to-end development and validation of predictive diagnostic solutions for motorcycles. This involved developing software that when supplied with data from the specific vehicle could analyse it and assert the vehicles health and predict failures. This effort involved making a testable platform of vehicle fleet, a communication infrastructure with custom protocols, a cloud driven data collection and analysis system and the solutions themselves. Our team worked with vehicles of different make, developed custom hardware for data collection and specifically wrote cloud-APIs and apps to analyze collected data and display it to the user.
I collaborated on a variety of projects within this connectivity project such as engine oil monitoring, vehicle battery monitoring and air filter monitoring, but personally led the development of two major solutions: sensor-less tire pressure monitoring and intelligent range estimation. These programs involved collecting time-series CAN data, utilizing statistical methods ranging from Kalman filters to Neural Networks to make accurate judgements about the state of the vehicle. These works gave rise to accurate monitoring of vehicle health and quantification of complex events such as driving style and aggression. This further led to four patents in this area, 3 published by July 2022 (Ref. 202241018611, 202141038618, 202141019724).
As a mechanical engineering student at National Institute of Techology Karnataka, my academic journey started with dabbling in dynamics & design and then gradually transitioning towards controls, system engineering and robotics. During my four years at NITK, I tried a lot of different projects while only diving deeply mostly into automotives and robotics. This includes an academic inclination towards intelligent systems, attending the system engineering conference INCOSE 2017 and an unsuccessful attempt at establishing a student satellite team at NITK while colloaborating with ISRO (Indian Space Research Organization). Every project taught me a little about engineering fundamentals and a lot about team management.
Final Thesis Project: For my final bachelors' thesis I and a team of three other students chose to work on a robotics topic. The advent of 3-D printing and small manipulators motivated us to develop and optimize a 3-PRRR manipulator (in the image above). This included design and optimization in both the hardware and the software. We managed to display strong results to our advisor with good position control and weight savings in the physical design.
Smart City Project: Another major project that I was happy to be a part of, was a smart city experiment, in which 50 students collaborated over 2 months to make a remotely controlled and automated smart city model with smart-buildings and vehicles with automated and prioritized functionalities. This project included mechatronic design, sensor fusion, wireless communication protocols, path planning, FMEA and writing efficient code.
Elected courses at a glance:
One of the most involving and rewarding experiences I had in college was my time with our 'Formula Student' team, NITKRacing. NITKRacing is NITK's answer to the Formula Student question: Can a bunch of students make a Formula One style single seater performance car? Based on a rigorous rulebook, students are challenged to design, manufacture and (mock) sell a weekend racecar at competitions around the world. Formula Student is not only a excrutiatingly intensive student project but also grounds for valuable research, innovative solutions and even top spot for recruitments.
I was a part of NITKRacing from 2016 to 2019 during which I participated in 3 events. In these events I was exposed to automotive engineering, industry experts, and invaluable team building experience. The design process encompassed everything from fancy FEM analysis for every manufactured component to fundamental mathematical models that the entire car was based on. Initially being heavily involved with part design such as pedals, uprights, wheel hubs, etc.; over time I slowly migrated towards higher lever system design, especially within vehicle dynamics. I was also elected captain of the team for year 2018-2019.
I was introduced to Indian Institute of Science as an intern trying to get some research experience. I had the the opportunity to work under the guidance of Prof. Satish Vasu Kailas on a short but interesting topic of designing in-situ testing rigs. Particularly I was asked to design a miniature fatigue testing rig and analze its failure modes for the SIAM lab for its experiments. Fatigue based design is a huge domain that spans mechanical design, civil engineering and structural engineering.
I started by first understanding the theoretical foundations of fatigue, writing down requirements, making prototype CAD models, analzing them in ANSYS and evaluating them against other designs. This meant designing to include available hardware like motors and sensors and predicting most crucial failure modes. I am very grateful for Prof. Kailas' guidance during the entire process.
During the winter of '17 I managed to squeeze a short but insightful internship at HAL Aircraft Division. This was a month-long stint that involved a thorough tour and deep-dive into various manufacturing and maintenance facilities at HAL's Aircraft Division at Bangalore. A group of students were exposed to various processes and materials and finally had to write an industrial report for our respective schools.
Sensor-Fusion and Position Estimation in IPhones: Utilizing camera feed + computer vision, GPS data and IMU data, we developed a system to determine position of other paired iPhones in a 3-D environment. This is done within a probablistic framework while quantifying uncertainty.