Computer Engineering Student · MKSSS Cummins College of Engineering, Pune
3rd Year · Building at the intersection of AI, Full-Stack & DevOps

• 🎓 Pursuing B.E. Computer Engineering at MKSSS Cummins College of Engineering, Pune · CGPA: 8.9/10
• 🏆 1st Place — Synapse 3.0 AI/ML Hackathon · 2nd Rank — LOOP CCEW BUFFER DSA Competition
• 💼 Currently interning at Barclays
• 🔭 Building across AI/ML, Full-Stack, Post-Quantum Cryptography, and DevOps
• 🌱 Exploring Deep Learning, LLM integrations, Cloud Infrastructure, and System Design
• 💡 I enjoy building end-to-end systems that solve real problems — not just demos
• 🧩 500+ problems solved on LeetCode

Languages

AI / ML

Web & Backend

DevOps & Tools

Databases

Full-stack AI waste classification platform · 🏆 1st Place, Synapse 3.0 Hackathon

• MobileNetV3 computer vision model (89.93% accuracy) + EasyOCR for plastic resin code detection
• FastAPI backend with JWT auth · React + Tailwind frontend with gamification (XP, badges, streaks, CO₂ tracking)
• Dual dashboards for citizens & municipalities · Groq AI chatbot · OpenStreetMap recycling centre finder
• Python PyTorch FastAPI React EasyOCR SQLite Groq AI

Real-time city power grid simulator · 🥈 2nd Rank, LOOP CCEW BUFFER DSA Competition

• 10 DSA components built from scratch: Min-Cost Max-Flow, Dijkstra, A*, BFS/DFS, Segment Tree, Fenwick Tree, Union-Find
• 14-node network with 4 zone types · fault injection, auto-reroute, and adaptive edge cost updates
• Live Swing-based visualisation with heatmaps, load bars, and sliding-window failure prediction engine
• Java Graph Algorithms Data Structures

Post-Quantum Cryptography Security Gateway for IoT Firmware Updates

• Bridges resource-constrained IoT devices to NIST FIPS 203/204 compliant quantum-safe cryptography (ML-KEM-768, ML-DSA-65)
• 4-detector ML anomaly ensemble (entropy, pattern, behavioral, size) scans firmware before crypto verification runs
• Crypto-agile live dashboard · full audit log · threat panel · solves Harvest-Now-Decrypt-Later (HNDL) attacks
• Next.js TypeScript Tailwind CSS @noble/post-quantum Recharts

Multi-model AI system for food recognition and personalised nutrition recommendations

• 3-model pipeline: CLIP zero-shot classifier → ViT multi-task DNN → DQN Reinforcement Learning agent for meal recommendations
• 62 dishes with Indian cuisine focus · Top-5 accuracy 85–88% · Calorie estimation within ±40–60 kcal
• 4 personalised health goal profiles: weight loss, muscle gain, diabetes management, maintenance
• Python PyTorch CLIP ViT DQN Gradio HuggingFace

End-to-end CI/CD pipeline with automated testing and containerised deployment

• Full pipeline: Git commit → GitHub Webhook → Jenkins → Maven build → Selenium UI tests → Docker deploy
• Structured branching strategy (feature → dev → main) with automated test validation before every production merge
• Java Spring Boot Jenkins Docker Selenium Maven

• 🥇 1st Place — Synapse 3.0 AI/ML Hackathon (National Level)
• 🥈 2nd Rank — LOOP CCEW BUFFER DSA Competition (College Level)
• 🧩 500+ problems solved on LeetCode
• 🎓 Honors in Advanced Machine Learning (Pursuing)
• 💼 Interning at Barclays

"Building real things, one commit at a time."