Somitra Singh Kushwah

SOMITRA SINGH KUSHWAH

*******************@*****.*** +1-682-***-**** linkedin.com/in/somitra-singh-kushwah EDUCATION

University of Texas at Arlington, Texas, USA 2024 - Present

● MS CS

Shri Vaishnav Vidyapeeth Viswavidyalaya, India 2019 - 2023

● B.Tech CS

WORK EXPERIENCE

Red Hill Labs Software Developer Jul’23 - Jul’24

● Engineered a centralized inventory management tool using Node.js and DynamoDB, consolidating sales, product, and stock data. This reduced manual data entry time by 40% and eliminated reconciliation errors.

● Implemented real-time data ingestion and reporting, improving inventory tracking accuracy by 15% and reducing the time to generate sales reports from hours to seconds..

● Championed Agile methodologies, leading daily stand-ups and sprint planning sessions that resulted in a 20% increase in on-time feature delivery for the inventory management project. Red Hill Labs Full Stack Developer Intern Jan’23 - May’23

● Built and deployed the Healthy Inside website using MERN stack (MongoDB, Express.js, React, Node.js) and modern HTML/CSS

● Conducted code reviews and participated in agile processes, increasing code quality by 20% and reducing bugs by 30%

● Deployed the Node.js web application on AWS Elastic Beanstalk and provisioned Amazon DynamoDB tables. PROJECTS

Hotel Booking Whatsapp AI Chatbot Dec’24

● Engineered a WhatsApp-based hotel booking chatbot using Flask and Twilio API, used OpenAI's API for natural language processing and function calling to streamline user interactions and query processing

● Demonstrated a significant reduction in manual intervention by automating the hotel booking workflow in simulated environments, resulting in faster response times and improved system efficiency.

● Integrated MongoDB database with the chatbot system to manage real-time room availability checks and automated booking confirmations, enhancing the hotel reservation workflow Landmark Detection Using Machine Learning Apr’25

● Conducted EDA on the 4M-image Google Landmark dataset, selecting the top 20 classes via IQR balancing.

● Developed and benchmarked multiple CNN architectures (CustomCNN, ResNet50, VGG19, and MobileNetV2), utilizing transfer learning and hyperparameter tuning to optimize feature extraction capabilities.

● Achieved a peak validation accuracy of 87.57% with MobileNetV2. Real-Time Hand Gesture-Based Virtual Keyboard and Mouse Apr’25

● Developed a real-time dynamic hand-gesture recognition system by extracting MediaPipe Hand keypoints and computing angular features to represent each frame.

● Designed and implemented a three-layer LSTM network in PyTorch to classify four distinct gestures (“Zoom In,” “Zoom Out,” “Swipe Up,” “Swipe Down”) from temporal keypoint sequences.

● Trained the model over 50 epochs, achieving ~84 % validation accuracy, and exported the final weights for integration into live gesture control applications.

SKILLS

Languages: Python, JavaScript, Java, C++, C, SQL

Libraries & Frameworks: React.js, Node.js, Express.js, PyTorch, Hugging Face, Langchain, LangGraph, Pandas, Scikit-Learn, OpenCV, Matplotlib

Developer Tools & Platforms: AWS (EC2, DynamoDB, Elastic Beanstalk), MS-Excel, Git, Postman, VSCode, Power BI, Tableau

Areas of Expertise: Machine Learning, Deep Learning (CNN, RNN), NLP, Computer Vision, REST API Development, Web Development

Contact this candidate