• Developed end-to-end post-trade report automation pipelines using Python and Bash, integrating task scheduling, file transfer (SFTP), and logging systems to generate, validate, and route 30+ daily reports to downstream systems.
• Automated Excel-based workflows and developed VBA-powered dashboards to monitor application health, enabling real-time detection of status target breaches and reducing manual oversight by 75%.

• Led the redesign and development of the company’s website using React, Next.js, and TailwindCSS, resulting in a 60% improvement in load speed and a fully responsive, SEO-optimized interface across all modern browsers.
• Developed a lightweight Node.js + Express backend service to scan inbound emails for malicious links, leveraging regex and domain blacklisting logic, blocking an estimated 250+ phishing attempts during the internship.

• Identified and successfully resolved a critical OpenCV rendering bug that previously disrupted drone image feeds, improving system uptime by 20% and ensuring reliable mission performance.
• Optimized image processing pipelines to resolve compatibility issues caused by mismatched GPU driver versions and OpenCV dependencies, reducing latency by 300 ms and improving frame processing efficiency by 25%.
• Designed automated image processing scripts using Python, improving obstacle detection accuracy by 40%.

• Developed over 50 programming lessons, adapting content to address the learning needs of 200+ students.
• Mentored students in advanced Python concepts, including object-oriented programming, data structures, and algorithm design, enabling them to write efficient and modular code for scalable software projects.

• Led the robotics team to secure 3rd place out of 60+ teams in a highly-competitive regional competition by engineering cost-effective solutions and optimizing team roles to enhance efficiency and performance.
• Mentored 25 students in Java through 10+ workshops, enhancing troubleshooting skills and optimizing robotic systems for seamless software-mechanical integration in high-stakes competitions.

• Developed a deep learning-based waste classification system using PyTorch that achieves 90% accuracy in distinguishing between organic and recyclable materials, deployed as an educational tool reaching 150+ elementary school students.
• Created a waste management solution with a web interface, incorporating a YOLO-based neural network that reduced classification errors by 40% compared to baseline models while processing 10,000+ diverse waste images for training.

• Engineered a full-stack TypeScript/React flashcard application with Supabase integration, implementing a spaced repetition system (SRS) using the FSRS algorithm with configurable parameters for optimized learning retention rates of 90%.
• Developed a real-time study analytics system using React hooks, tracking user progress, statistics, and learning streaks.

• Developed a full-stack habit tracking app with a PostgreSQL database, integrating user-initiated financial pledges, using Next.js, TypeScript, Tailwind CSS, and Supabase.
• Built a user-friendly app with a modern UI, fostering accountability and charitable giving through user pledges.

• Built a full-stack multiplayer adaptation of the Spyfall game utilizing React and TypeScript, achieving a highly responsive user interface that facilitated over 1,000 game sessions within the first month of launch.
• Optimized Firebase backend for 20% faster latency and improved game state synchronization through data indexing.

• Developed a Python gas storage contract pricing model, achieving 82% accuracy compared to production models.
• Developed a Random Forest Classifier for loan default prediction, training on 10,000-entry dataset for accuracy.

• Built a visualization tool for a 3-layer, 4-neuron network, making machine learning concepts more accessible.
• Developed an interface for real-time parameter tuning and activation function visualization.