Mar 2023 – Aug 2024 · 1 yr 5 mos

• As a Lead Engineer in Machine Learning, I drove impactful innovations in machine learning and performance optimization to enhance user experience and system efficiency.
• Developed an Automated Benchmarking Tool:
• Led the design and implementation of a Python-based Android profiling tool, streamlining the process of benchmarking rendering across devices and identifying performance bottlenecks like scroll janks.
• Personalization Solutions with Edge-Based LLMs:
• Spearheaded the creation of four edge-based personalization solutions utilizing large language models (LLMs), leveraging data insights to elevate user experiences and optimize app interactions. Successfully fine-tuned the FLAN-T5 LLM to meet the unique needs of our application. Also collaborated with cross-functional teams to improve the dataset quality of the dataset, driving significant improvements in the accuracy and reliability of personalization solutions.

Mar 2021 – Feb 2023 · 1 yr 11 mos

• As a Senior Software Engineer in Machine Learning, I played a key role in building cutting-edge solutions that focused on user experience, privacy, and real-time performance.
• Edge ML Solution for Real-Time Boredom Detection:
• Designed and implemented an edge machine learning solution that analyzed phone usage data to detect boredom. Developed an Android app to deploy the model for real-time inference with under 50ms latency, significantly enhancing the user experience with context-aware features.
• Federated Learning for Privacy:
• Pioneered a Federated Learning-based solution to predict gender and demographic age, ensuring user data privacy. This approach aimed at minimizing data sharing while maintaining predictive accuracy.
• Explored Distributed Learning Techniques:
• Conducted research on and tested various Federated Learning algorithms and configurations to optimize the demographic age and gender prediction model, laying the foundation for future privacy-preserving AI advancements.

Jun 2019 – Feb 2021 · 1 yr 8 mos

• As a Software Engineer, I contributed to innovative solutions in edge ML and real-time data visualization, working with cross-functional teams to create impactful applications and publish research.
• Robot Camera Visualization Android App:
• Developed an Android app for real-time visualization of depth maps and 3D point clouds from a Time-of-Flight (ToF) camera, achieving 30 FPS with gesture-based UI features for enhanced user interaction. The app was tailored to meet stakeholder requirements and provide intuitive controls for visualizing complex data.
• Privacy-Preserving Edge ML Solution:
• Engineered an edge ML solution for the Next App Recommendation system, focusing on privacy preservation. Designed a memory-efficient model with a 99% size reduction, minimizing Federated Learning (FL) bandwidth costs. The model was trained and deployed in Java using DL4J, integrated into an Android edge device app (User Trial app) across 500+ devices for training and inference. This solution was published in IEEE ICSC 2022.

May 2018 – Jul 2018 · 2 mos · Bengaluru East, Karnataka, India

• As a Summer Intern, I designed and developed machine learning models to predict user behavior and connectivity, contributing to innovative research in mobile networks and location prediction.
• Neural Network for Location Prediction:
• Designed and developed a neural network (NN) model to predict a user’s location based on recent travel data and time of day, leveraging machine learning to improve location-based services.
• Simulation Environment for Cell Tower Connectivity:
• Created a Python-based simulation environment to model cell tower connectivity as users traveled, helping to simulate and analyze real-world mobile network behavior.
• ML Classification Model for Cell Tower Prediction:
• Built a machine learning classification model to predict the most likely cell tower a user would be connected to at any given time. Achieved Top-1 and Top-3 accuracies of 85-90% and 90-95%, respectively, on the in-house evaluation dataset.

Oct 2017 – Mar 2018 · 5 mos · IIT Kanpur

• As the Event Manager for Robogames in Techkriti 2018 at IIT Kanpur, I led the design, management, and execution of multiple robotics competitions, contributing to the success of one of the largest technical festivals in India.
• Event Design & Management:
• Played an integral role in conceptualizing and managing the various Robogames events and competitions during Techkriti'18, overseeing everything from event planning to logistics and team coordination.
• Successful Conduction of IARC: Led the successful conduction of the IARC at Techkriti'18, a flagship competition, ensuring smooth operations and high engagement from participants.

May 2017 – Jul 2017 · 2 mos · Delhi Aerocity, Delhi, India

• During my internship at Hike Pvt. Ltd., I focused on implementing machine learning models for image classification and deploying them for real-time predictions, utilizing cloud services for scalability and efficiency.
• CNN Implementation for Image Classification:
• Developed Convolutional Neural Network (CNN) models in Python using TensorFlow for image classification tasks, applying deep learning techniques to analyze visual data.
• Cloud-Based Model Training & Deployment:
• Leveraged Google ML Engine APIs to accelerate training on Google Cloud, ensuring efficient model training at scale. Deployed the trained model using TensorFlow Serving, exposing REST APIs to generate model predictions in real time.