
In My Own Words
I'm a Machine Learning Engineer with deep expertise in building and deploying cutting-edge computer vision systems for real-world applications — from autonomous trucks to threat detection and industrial automation.
Currently at Torc Robotics, I develop advanced perception systems for self-driving trucks, leveraging multi-modal sensor data (camera + LiDAR) and state-of-the-art deep learning techniques. My work spans supervised and self-supervised learning, attention-based models, and large-scale distributed training on cloud infrastructure. I've also worked extensively on monocular, stereo, and multi-view depth perception, delivering reliable dense depth outputs in real time.
I’ve built end-to-end ML pipelines — from data selection using active learning, to cloud-native model training and deployment on edge devices using TensorRT. My contributions include scalable infrastructure, robust error handling, and continuous improvement of models in production.
Previously, at Liberty Defense Technologies, I helped take a synthetic aperture radar-based weapons detection system from R&D to production, developing real-time object detection and semantic segmentation models. I worked across the stack: model design, training at scale, performance optimization, and CI/CD deployment in a startup environment.
Throughout my career, I've combined strong software engineering (Python, C++, Docker) with deep ML domain knowledge to solve hard perception problems. Whether it’s working with multi-sensor data, training custom CNNs/Tranformers, or deploying ML systems at scale, I thrive at the intersection of research and engineering.
Work Experience
SOFTWARE ENGINEER (MACHINE LEARNING)
LIBERTY DEFENSE TECHNOLOGIES2018 – 2020
• Working in a fast-paced startup environment with a small team. • Developed POC for a real time object detection and classification system using 3D, 2.5D and 2D images collected from Synthetic Aperture Radar. • Worked on building an Anomaly Detection model. • Collaborate with the RF team to find the best parameters for the imaging system using Agile techniques. • Trained interns to use data annotation tools in order to perform inhouse data labeling. • Deploy Object detection and classification model using docker containers. • Define unique set of error metrics exclusive to our system and automated the process of evaluation. • 3D data filtering using Depth masking, thresholding and filtering. • Tools and Languages: C++, python, Matlab, bash scripts (basics), Docker (basics), OpenCV. • Models used: CNN, Auto encoder, SVM, one class SVM, KNN, XGboost. dimensionality reduction (PCA, ICA), Ensemble learning, Wavelet Transform.
RESEARCH AND DEVELOPMENT INTERN
UHV TECHNOLOGIES2018 – 2018
FORTH WORTH, TX • Worked on real time scrap metal separator that used computer vision and real time object classification model to segement different types of scrap metal. • Captured 2D Images of scrap metals to train the Image classifier model. • Performed object detection, Segmentation, Tracking and classification in real time. • Train, validate and test the image classifier model using various error metrics. • Tools and languages: C++, python. • Models used: CNN (Inception V3).
Education
BACHELOR OF ENGINEERING
OSMANIA UNIVERSITY2016
BACHELOR OF ENGINEERING
MASTER OF SCIENCE
UNIVERSITY OF TEXAS AT ARLINGTON2018
MASTER OF SCIENCE