Resume/CV

Education

Indiana University Bloomington
May 2024
Master of Science in Computer Science CGPA: 3.88/4
Courses: Applied Algorithms, Deep Learning Systems, Elements of Artificial Intelligence, Applied Machine Learning, Software Engineering, Computer Networks

ICFAI Foundation for Higher Education
Jun 2020
Bachelor of Technology in Electronics and Communication Engineering CGPA: 10/10

Skills

Technical Skills: Python, Java, C++, SQL (MySQL), HTML, CSS, Git, Jupyter, Flask, AWS (EC2, S3), GCP, Docker

Libraries and Frameworks: Scikit-Learn, TensorFlow-Keras, PyTorch, PyTorch Lightning, OpenCV, ONNX, Apache TVM, TensorRT, CoreML, Flask.

Experience

AiKYNETIX
May 2023 - Present
Machine Learning Engineer
Houston, Texas, USA

• Ported state-of-the-art human pose estimation models from PyTorch to CoreML format for AiKYNETIX’s human motion video analytics iPhone application, improving performance of the app by 25%
• Collaborated with the web development team and orchestrated the seamless launch of AIK-Web, AiKYNETIX’s cutting-edge web-based platform for human motion analytics, with a 20% increase in the accuracy of the metrics
• Deployed the web application end-to-end using Docker with support for switching different pose estimation models
• Explored PyTorch 2.0’s compile and TorchDynamo enabling faster and easier conversion of PyTorch models to CoreML

IU Computer Vision Lab
May 2023 - Present
Research Assistant
Bloomington, Indiana, USA

• Engineered novel diffusion-based depth map refinement and super-resolution networks to enhance depth map quality in multi-view stereo networks, optimizing 3D scene reconstruction outcomes
• Researched 20+ papers based on attention and diffusion-based networks for depth-map super-resolution

Cyberinfrastructure for Network Science Center (CNS)
Oct 2022 – Apr 2023
Research Assistant - Machine Learning Bloomington, Indiana, USA

• Collaborated with the Nolan Lab at Stanford and the Van Valen Lab at CalTech in performing single-cell segmentation and cell neighborhood analysis on CODEX multiplexed imaging data from 7 organs and integrating the pipeline into HuBMAP’s Human Reference Atlas.

Segmind
Apr 2022 – Jul 2022
Computer Vision (CV) Consultant Hyderabad, Telangana, India

• VoltaML
  • Developed a lightweight library to accelerate AI, ML, and DL models by optimizing, compiling, and deploying to target CPU and GPU devices.
  • Accelerated model inference by up to 10x compared to native PyTorch by leveraging ONNX, TorchScript, TensorRT, and Apache TVM
  • Incorporated pipelines for accelerating image classification, object detection, semantic segmentation, and natural language processing models.
• Object Detection in Aerial Drone Videos
  • Developed an object detection tool to detect minuscule objects (<5x5 pixels) in aerial drone videos.
  • Achieved an mAP of 0.72 and reduced manual annotation effort by 50%.

Onward Assist
Jan 2020 – May 2022
Machine Learning Engineer Hyderabad, Telangana, India

• MoNuSAC: Multi-Organ Nuclei Segmentation and Nuclei Classification
  • Pioneered U-HoVerNet, a network to segment cells in regions of interest (ROI) in whole slide images, scoring a Panoptic Quality (PQ) of 0.39.
  • Secured 6th place in the post-challenge leaderboard of the MoNuSAC 2020 competition.
  • Co-authored the MoNuSAC2020: A Multi-organ Nuclei Segmentation and Classification Challenge paper, published in IEEE TMI.
• Lung Nodule Detection in Chest Radiographs
  • Implemented FasterRCNN, and RetinaNet models to classify and detect cancerous lung nodules in chest X-Rays.
  • Attained an AUROC of 0.83 for nodule classification and mean IoU of 0.76 for nodule detection after fine-tuning and hyperparameter search.
• Real-time Medical Image Stitching
  • Implemented an end-to-end medical image stitcher using OpenCV that stitches frames captured from a camera-mounted microscope.
  • Facilitated real-time stitching of frames with exceptional quality, by propelling stitching speed to more than 10 frames per second.
  • Created a desktop application for the tool using PyQt5 and integrated the app into TelePathDx, a digital pathology platform by Onward.

Projects

Equitable Data Valuation for Image Segmentation
Feb 2021 - Dec 2021

• Applied KNN-Shapley, a metric used to determine the quality of data based on the training algorithm used, to image segmentation algorithms.
• Detected 80% of poor-quality data by inspecting the lowest 20% of KNN-Shapley valued training data points, validating the research approach.

Paper Reproduction
Apr 2022

• Reproduced 2 papers viz., “Image Style Transfer using Convolutional Neural Networks” and “Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks”.