cv

Masters of Science in Computer Science (CGPA - 4.0 / 4.0)

University of Colorado Boulder

Bachelors of Technology in Electronics and Communication Engineering

Indian Institute of Technology (IIT) Guwahati, India

RMap Millimiter-Wave Radar Mapping through Volumetric Upsampling (IROS 2024)

Tell Me Where to Go A Composable Framework for Context-Aware Embodied Robot Navigation (CoRL 2023)

CryoSegment Simultaneous Segmentation of diverse cellualr structures from Cryo-ET images (under review for Nature Methods)

Grouped BERT for Multi-Label classification to reason the human values behind the arguments (Sem-EVal 2023)

Research Assistant

Vignesh Kasinath Lab

• Designed end-to-end pipeline for semantic segmentation of low resolution and sparse cellular structures from cyro electron tomography images using Multi-UNet architecture, resulting in a significant 13% boost in F1-score
• Developed a 2-stage semi-supervised segmentation framework to segment cellular regions and fine-grained cellular structures, reducing the number of manual annotations by 85%
• Developed unsupervised algorithms to effectively segment different instances of chromosomes in a nucleus
• Designed advanced image processing pipelines for precise 3D pose extraction of cellualr structures from volumetric microscopic imaging data, leveraging connected components, spline interpolation, and geometric transformations
• Integrated fine-tuned Segment Anything Model (SAM) in the pipeline, eliminating manual parameter tuning and enhancing the performance of the segmentation masks by 20%

Research Assistant

Autonomous Robotics and Perception Group

• Developed RMap, a volumetric mapping framework that transforms sparse radar pointclouds to high resolution 3D lidar-like maps, achieving a significant improvement in F1-score 0.59 compared to 0.19 of traditional radar maps
• Designed UpPoinTr transformer for enhancing volumetric maps, surpassing the prior models by 8% in F1-score
• Developed the pose-based region sampling algorithm to efficiently represent the 3D maps as minimal partial pointclouds, without any loss in details
• Developed custom CUDA pointcloud loss metrics to evaluate deviations from groundtruth in all three dimensions
• Enhanced the odometry estimation by 8% using only the mmWave radar, by extracting the features from 3D radar images using transformer encoder to predict a 6DoF pose

Computer Vision Research Engineer

Samsung Semiconductord India R&D

• Developed real-time 3D scene reconstruction algorithm, only using depth from ToF sensors, optimized to 20fps
• Improved the accuracy by 5% of the reconstructed scene by removing outliers using gaussian smoothing
• Automated pose alignment of reconstructed 3D scene with groundtruth, eliminating manual alignment in MeshLab
• Reduced the latency of Remosaic deep learning models for 200M pixel camera sensor using quantization and pruning techniques by 10% with an unnoticeable degradation of 0.1% in perceptual quality
• Optimized Samsung CMOS camera sensor noise reduction algorithm with OpenMP parallel programming, thereby reducing the algorithm runtime by 33%

Software Engineer

Qualcomm

• Spearheaded the development and maintenance of python audio library to evaluate both the objective and perceptual audio quality of Bluetooth headsets
• Enhanced python automated test framework with new features that populate test vectors and visualize audio output signals, leading to a 10%-15% reduction in both the validation and developments efforts

Research Intern

Eagle Eye Networks

• Predicted the pedestrian/car trajectory in a video by estimating the optical flow for the YOLO precited objects
• Achieved 72% accuracy in segmenting location and trajectory anomalies through Incremental Spherical Clustering

• Received Employer of the Month at Samsung, for successfully optimizing the noise reduction algorithm to reduce runtime by 33%

• Received the Spot Appreciation award, for exceptional performance in delivering the 3D reconstruction pipeline for Samsung's ToF sensors

• Received Qualstar from Qualcomm for continued excellence in improving the audio quality analysis framework