I am a 4th-year B.Tech student at IIIT Allahabad . As a Full Stack Developer and Machine Learning enthusiast, I focus on building scalable applications and optimizing deep learning models. I specialize in architecting responsive web solutions while actively researching transformer-based architectures for predictive modeling.

• Development Stack: I build high-performance web apps using React, Node.js, and PostgreSQL, and architect serverless backends using AWS, Vercel, and Redis.
• Machine Learning & AI: My focus is on fine-tuning LLMs, leveraging tools like PyTorch, TensorFlow, and HuggingFace, alongside optimizing data pipelines
• Algorithmic Problem Solving: I am an Expert on Codeforces with a maximum rating of 1657. I have solved over 600 problems on LeetCode and 150+ problems on GeeksforGeeks, ensuring efficient and optimized code in production environments.

• Amazon ML Challenge 2025 - Achieved AIR 17 among 3,000+ teams nationwide.
• Secured a Top 100 finish at the Meta PyTorch OpenEnv Hackathon x Scaler School of Technology 2026 .
• Amazon ML Summer School 2026 - Selected among 3,000 participants from over 130,000 applicants nationwide.
• Flipkart Runway Season 5 - Selected among the Top 100 from over 46,000+ participants nationwide .
• Meesho Scripted By Her 2.0 Hackathon - Selected among the Top 500 from over 46,000+ participants.
• Selected for Google Student Launchpad 2025, a Google-led initiative for technical upskilling.
• Expert on Codeforces with Max. Rating: 1657 ( Profile)
• Solved 600+ problems on LeetCode ( Profile) and 150+ problems on GeeksforGeeks ( Profile).
• NTSE Scholar - Qualified both Stage I and Stage II of the National Talent Search Examination.

• Jira-To-Code Agent Environment: Built an autonomous framework using Python and FastAPI for LLMs to resolve Jira tickets via sandboxed workspace edits. Engineered isolated Docker runtime environments featuring deterministic Pytest validation pipelines to provide verifiable rewards.
• QuizPulse: Developed a production-ready, full-stack multiplayer quiz platform using React, TypeScript, Node.js, and MongoDB. Integrated the Google Gemini API to generate quizzes from user-provided topics and PDFs using AI-driven question generation.
• CorpGym (Corporate Shared Workspaces): Architected an enterprise planning environment modeling multi-agent task orchestration using Python and Reinforcement Learning. Formulated a composite reward signal and evaluated a Qwen 2.5 Base SFT RLVR pipeline, achieving a 100% success rate on multi-turn budget reallocation tasks.
• LeetCodeStats Dashboard: Created a serverless full-stack web application using Node.js and Vercel to fetch and visualize real-time user profile statistics via the official LeetCode GraphQL API. Architected a cache-first backend using Upstash Redis to securely bypass CORS restrictions and reduce redundant API requests.
• ClauseRisk Analyzer: Engineered modular prompt execution pipelines using Python, LLMs, and Streamlit to systematically traverse deep document layouts and semantically isolate high-risk operational text clauses.
• Product Price Prediction Engine: Cleaned a large-scale noisy dataset of 150,000 product listings and trained a transformer-based regression model to capture non-linear pricing patterns. Integrated Phi-3 Mini for zero-shot pattern parsing.