Hi, I'm Urva Gandhi, a Computer Science undergraduate at Nirma University. I'm a passionate software engineer who thrives at the intersection of backend architecture and artificial intelligence. My journey involves transforming complex ideas into elegant, production-ready solutions.
Currently, I'm building CoinTrack (Unified financial analytics) and CodeGuardian (AI-powered secure code analysis). I'm deep diving into Data Structures & Algorithms mastery, LLM Application Development & RAG systems, and High-Performance Backend & System Design.
Open to collaborating on real-time financial data pipelines, ML deployment workflows & MLOps, and turning Hackathon prototypes into Production systems. I've solved 200+ LeetCode problems and completed 4+ major projects.
Built a healthcare analytics platform predicting hospital readmissions (30/60/90 days) and delivering cost + clinical insights. Designed a Smart Model Loader for instant evaluation of ML models (Logistic Regression, Random Forest, XGBoost, Deep Learning). Developed an interactive Streamlit dashboard with visualizations.
Participated in Adobe's 'Connecting the Dots' hackathon. Built an offline PDF Outline Extractor using heuristics on font size, boldness, and layout. Designed a persona-driven document intelligence pipeline with keyword filtering + semantic ranking. Optimized processing: <10s per 50-page PDF.
Participated with innovative AI-driven solutions for real-world problems. Gained experience in rapid prototyping and collaborative development under time constraints.
Minor in Adaptive AI. CGPA: 8.79. Relevant Coursework: Machine Learning, Deep Learning. Active participant in hackathons and coding competitions.
Percentile: 99.28. Focused on Science stream with Mathematics and Computer Science.
Percentile: 96.92. Strong foundation in Mathematics and Science.
Unified finance dashboard aggregating portfolio data from multiple stock broker APIs (Zerodha, Angel One, etc.) into a single view. Features P&L tracking, live market data, watchlist, and exportable reports.
Built a platform predicting hospital readmissions (30/60/90 days) with Smart Model Loader for instant ML evaluation. Features interactive Streamlit dashboard with visualizations.
Offline PDF analysis engine built for Adobe Hackathon. Features structured outline extraction with hierarchy detection, and persona-driven document intelligence that adapts content for different user roles.
AI-driven multi-language vulnerability detection engine using graph-aware transformers and static analysis. Auto-scans code to identify security flaws, maps to CWE/CVE, and generates explainable remediation suggestions.
