I'm a Computer Engineering student specializing in AI/ML, with a passion for building end-to-end systems.
Final-year Computer Engineering student at EPCET Bengaluru, specializing in Artificial Intelligence and Machine Learning, with a GPA of 8.8/10.
I work at the intersection of AI, machine learning, and scalable software engineering — transforming complex data into systems that are practical, resilient, and built for real-world impact. My focus goes beyond training accurate models; I'm driven by designing intelligent solutions that remain reliable under uncertainty, adapt to imperfect data, and perform consistently in production environments.
I compete in hackathons to stress-test my speed and adaptability, and I thrive on learning new tools and techniques that push the boundaries of what's possible with data.
Real pipelines, real data, real evaluation. Not tutorials — systems engineered to perform under pressure.
A production-grade AI security platform that predicts zero-day vulnerabilities in source code before they're exploited. Built under an intense 8-hour hackathon constraint, ZeroVA combines a fine-tuned CodeBERT transformer for semantic code understanding, a Graph Neural Network to model inter-function dependencies, and a 4-model ensemble (XGBoost, LightGBM, CatBoost, Random Forest) trained on 8,000 CVE-annotated samples. The system is fully deployed via FastAPI on Hugging Face Spaces with Docker.
A full-stack AI writing assistant powered by Meta's LLaMA 3.3 70B, designed to generate contextually aware content across 8 distinct tones, 7 audience profiles, and 7 output formats — from technical documentation to persuasive copy. Quillox includes a production-grade AI safety layer with prompt injection detection and jailbreak filtering to prevent misuse, plus a daily token quota system to manage inference costs. Deployed on Render with GitHub auto-deploy for continuous delivery.
Architected a classification pipeline across 95,000+ URLs using lexical and statistical feature extraction with Python and scikit-learn. Benchmarked six models end-to-end; Random Forest achieved a 0.96 F1-score while ensemble stacking reached 0.91 macro F1 on completely unseen data. Applied SMOTE to correct severe class imbalance and maximize zero-day phishing detection.
Won 2nd place among 100+ competing teams by building ZeroVA AI — a production-grade security platform combining CodeBERT, a Graph Neural Network, and a 4-model ML ensemble that achieved 97.3% accuracy on CVE-annotated code samples.
Competed in the College Track of the USAII® Global AI Hackathon 2026, placing 286th out of 424 teams among over 6,000 global participants, demonstrating applied AI/ML skills on an international stage.
Developed an AI-driven productivity solution using Generative AI for workflow automation and task optimisation. Shortlisted in the national top 100, competing against working industry professionals.
Selected among 1,000+ global participants. Built an AI-based mental health application leveraging machine learning for personalised user recommendations.
Advanced to Top 5,000 out of 25,000+ teams globally. Developed a rapid AI/ML prototype under 8-hour constraints, focusing on real-time problem solving.
Final Round qualifier in a national-level coding competition, demonstrating strong problem-solving skills using data structures and algorithms.
Tools and techniques I use to build, evaluate, and deploy machine learning systems.
Actively seeking AI/ML engineering roles, research collaborations, and internship opportunities. Whether you have a model to train or a pipeline to build — let's talk.