Full-Stack · AI/ML · Cybersecurity
Pre-final year B.Tech CSE (Information Security) at VIT Vellore.
I'm a Computer Science student at VIT Vellore with a passion for building systems that sit at the intersection of full-stack development, artificial intelligence, and cybersecurity.
From designing deepfake detection systems to deploying IoT health monitors, I love turning complex problems into elegant, production-ready solutions. When I'm not coding, I'm contributing to the security community on TryHackMe or mentoring peers in technical clubs.
B.Tech CSE (Info Security) — VIT Vellore
Vellore, Tamil Nadu, India
Full-Stack · AI/ML · Cybersecurity
B.Tech in Computer Science & Engineering (Specialization: Information Security)
Relevant Coursework
A versatile toolkit spanning full-stack development, AI, and cybersecurity.
From AI models to cybersecurity labs — real projects solving real problems.
Full-stack dashboard integrating real-time stock prices & news from 5+ APIs (Alpha Vantage, NewsAPI). AI-driven sentiment analysis pipeline using Google Gemini API, reducing manual effort by 45%. Dynamic interactive visualizations with Recharts improving trend-analysis efficiency by 35%.
End-to-end deepfake detection system using EfficientNetB4 CNN for real vs. fake classification. Flask REST API with YouTube integration, frame-by-frame confidence scoring. Chrome extension for instant social media verification.
Reconstructs a 10-sec 1080p 30FPS video from 300 shuffled frames. Hybrid similarity with ResNet50 deep embeddings + SSIM. Greedy + TSP-inspired 2-opt optimization achieves 97–99% accuracy, 40% faster via parallelized extraction.
Full-stack app where users input raw text feelings; Gemini AI analyzes emotion, Last.fm fetches songs, and an AI-curated 10-song playlist structured in 3 acts is generated with YouTube embedded players for inline playback.
Multi-party video calling in modern C++20 built on a custom RTP/SFU stack instead of libwebrtc — the protocol work is the point. 1:1 and up to 8-party calls via a custom Selective Forwarding Unit, real cross-laptop camera + mic over a custom ICE/RTP transport (libjuice), H.264 (FFmpeg/x264) video and Opus audio, targeting <150 ms glass-to-glass latency with RTCP REMB adaptive bitrate. 10/10 unit tests; ~5,374 eight-party rooms/core.
Blockchain-secured audio-visual authentication combining 5-pillar multi-modal deep learning with cryptographic provenance. Analyzes video through five independent forensic perspectives (473-dim fingerprints) with per-pillar explainability; ensemble reaches 81.7% accuracy / 0.877 AUC-ROC on the LAV-DF dataset, with a transfer-learning branch at 92%. SHA-256 provenance chains create tamper-proof audit trails.
Hands-on web penetration-testing lab against a deliberately vulnerable Flask application. Exploited and documented OWASP-style vulnerabilities — SQL injection, XSS, SSTI, command injection, insecure deserialization, broken authentication, file upload, and information disclosure — building practical exploitation and remediation skills in a safe, educational environment.
Modular health monitoring system with Python-based vital signs simulator (heart rate, SpO₂, BP, temperature), FastAPI server for data processing, and real-time web dashboard for visualization.
End-to-end ML pipeline with Random Forest Classifier predicting Titanic passenger survival. Interactive Streamlit web app for real-time predictions. Dynamic 3D Plotly visualizations exploring age, class, and gender relationships.
SOC Level 1 (65 hrs): Alert triage, SIEM fundamentals, log analysis, incident response, suspicious network traffic investigation. Jr Penetration Tester (30 hrs): Reconnaissance, scanning, exploitation, post-exploitation. Top 4% in India.
Scalable social media platform with authentication, authorization, audio tweets, forgot-password flow, subscription payments, and Vercel deployment. Built during NullClass EdTech Software Engineering internship.
Production-quality pipeline over 20 Newsgroups (~20K posts). 6-stage cleaning pipeline, 384-dim L2-normalized embeddings in ChromaDB. GMM fuzzy clustering with BIC/AIC selection. Hand-built semantic cache at O(n/k), 100% precision, 36% hit rate. FastAPI REST service.
Filed inventions in AI-driven cybersecurity and cryptographic access control.
Deep learning and adaptive policy networks for real-time network security. A five-model AI framework — Bidirectional LSTM behavioral learning, Autoencoder + Isolation Forest anomaly detection, an LSTM threat predictor, a Deep Q-Network for adaptive self-learning policies, and a multi-layer correlation engine — detecting IP, MAC, ARP, DNS spoofing and DDoS attacks at 98.7% accuracy with a 1.1% false-positive rate.
A machine-learning-enhanced access control and participant-selection layer for threshold cryptographic operations and secret sharing. Sitting above unchanged primitives like Shamir's Secret Sharing, it uses multi-dimensional trust vectors, ML-based trust-degradation prediction, and graph-based (PageRank-style) collusion pre-screening to decide who may participate — with zero-knowledge policy-compliance verification and full participant lifecycle management.
Building intelligent systems that solve real-world challenges.
CNNs, LSTMs, ensemble models, and transformer architectures with TensorFlow, Keras, and Scikit-learn.
End-to-end pipelines: preprocessing, feature engineering, training, tuning, and production deployment.
OpenCV, EfficientNetB4, ResNet50 for object detection, image classification, deepfake detection, and video analysis.
Text analysis, semantic search, and LLM integration using Sentence Transformers, ChromaDB, and Gemini API.
Offensive and defensive security specialization.
Network scanning (Nmap), packet analysis (Wireshark), and firewall configuration for enterprise environments.
Burp Suite, Metasploit, and OWASP methodologies for web application security testing.
SIEM log analysis with Splunk, incident response, threat hunting, and security event correlation.
Evidence collection, chain-of-custody procedures, and forensic analysis for incident investigation.
Research Intern
Software Engineering Intern
Web Development Intern / Software Engineer Intern
Led large-scale events with 20+ members achieving 100% on-time delivery. Spearheaded full-stack development for club projects (registration portals). Improved participation efficiency by 30%. Mentored junior members.
Spearheaded planning and execution of major club events. Led and motivated teams by delegating responsibilities and setting clear goals. Oversaw logistics, budgeting, and stakeholder communication.
Ranked top 4–5% on TryHackMe in India leaderboard
Completed SOC Level 1 (65 hrs) & Jr Pen Tester (30 hrs)
Solved 250+ data structures & algorithms problems on LeetCode
Active open-source contributor with 15+ projects on GitHub
100% on-time delivery for all large-scale club events at VIT Vellore
Have a project in mind? Let's build something great together.