Experience

• 

  National informatics centre (nic), india

  May 2022 - Jul 2022

  Data analysis and optimizations engineer

  1. Optimization of PM-KISAN Website: • Website and Server Optimization: i) Enhanced the overall performance of the PM-KISAN (Pradhan Mantri Kisan Samman Nidhi) website. ii) Optimized server configurations and resource allocation to improve the website's responsiveness and reliability. iii) Refined the underlying algorithms to ensure efficient data processing and user interaction.2. Development and Customization of De-duplication Algorithm: • Algorithm Development: i) Developed a robust algorithm to address the issue of data de-duplication. ii) Ensured each farmer was provided with a unique ID, eliminating redundancies and improving data integrity. • Performance Enhancement: i) Customized the algorithm to significantly increase the speed of search queries. ii) Implemented measures to enhance the accuracy and efficiency of data retrieval processes, resulting in a more streamlined and effective system. Show less

• 

  Indian institute of information technology allahabad

  Jun 2022 - Aug 2022

  Research and analyst intern

  1. Reconnaissance Gathering from Email: • Automation of Recon Gathering: i) Automated the process of reconnaissance gathering from emails using advanced techniques such as Google dorks. ii) Utilized open-source tools like OCCRP, Yasni, TinEye, and others to enhance data collection and analysis. iii) Streamlined the workflow to ensure efficient and accurate intelligence gathering from various email sources.2. Hate Speech Detection using Machine Learning: • Development of Multi-lingual Model: i) Developed a machine learning model capable of detecting hate speech across multiple languages. ii) Created a custom corpus and integrated various features to facilitate the detection of hate speech in both English and Hindi. • Real-world Data Testing: i) Scraped tweets from Twitter (now X) to compile a dataset for model testing. ii) Classified tweets based on hate speech with an accuracy of 95%. iii) Ensured the model's effectiveness and reliability in identifying and categorizing hate speech in practical scenarios. Show less

• 

  Tvs motor company

  May 2023 - Jul 2023

  Research and development intern - machine learning

  Designed and developed a robust backend Flask server to facilitate voice-activated automobile command execution, simulating real-world user interactions through voice alone. Key aspects of the project include: 1. Voice-activated Command Execution: • Created a system where users can perform various automobile commands using only their voice, enhancing hands-free usability and safety. 2. User Validation: • Developed a user validation mechanism leveraging a Gaussian Mixture Model (GMM). • Collected real-time voice data with environmental noise to train the model, achieving an accuracy rate of 85%. 3. Command Processing: • Implemented intents identification methods to accurately interpret and respond to diverse bike-related commands. • Ensured the system can handle a wide range of commands with precision, providing appropriate and reliable responses. 4. Real-world Application: • Focused on creating a practical and user-friendly solution for voice-activated automobile control. • Emphasized the importance of accuracy and reliability in user validation and command execution to enhance the overall user experience.This project demonstrates a blend of innovative machine learning applications and practical software development, aimed at revolutionizing how users interact with their vehicles using voice commands. Show less

• 

  Tvs motor company

  Jan 2024 - May 2024

  Dissertation project - r&d intern - ml developer

  1) Objective: - Developed an authentication system utilizing audio input to replace traditional input methods with voice commands. - Enhanced road safety by enabling riders to perform tasks without diverting attention from the road.2) Key Features: • Speaker Recognition: - Identified users based on their voice to ensure secure login and prevent unauthorized access. - Used the ECAPA-TDNN model from SpeechBrain, fine-tuned with a custom dataset for precise user identification. • Speech Recognition: - Classified commands into three categories: Updates (Check, Perform, Cancel), Edge (Battery status, Tyre pressure details, Fuel status), Cloud (Weather, Traffic, Notifications, Songs, News). - Concurrent threads for user identification and command classification (SRSR - Speaker Recognition during Speech Recognition).3) Technological Implementation: - Utilized a pre-trained Large Language Model (LLM) and fine-tuned hidden layers for optimal performance. - Achieved user voice identification with 20 seconds of training and 10 seconds for prediction. - Hosted the system using Flask with route-based functions for seamless integration and testing.4) Deployment and Testing: - Developed a front-end application with a simple UI for demonstration. - Planned integration into TVS vehicles' smart dashboards upon successful end-to-end testing. Show less

• 

  University of michigan-dearborn

  Oct 2024 - now

  As a Student Program Assistant, I support students by connecting them with essential resources, including food assistance, professional clothing, hygiene products, and emergency device rentals. I help manage the Student Food Pantry and Wolverine Wardrobe, coordinate outreach efforts, and assist with events in collaboration with community partners. Additionally, I provide guidance to students facing challenges such as housing insecurity, financial emergencies, and health concerns, ensuring they receive the support they need to succeed. Show less

  • Research Assistant

    Dec 2024 - now

  • Student Program Assistant at Dearborn Support

    Oct 2024 - now