Final-year CS undergrad specialising in Data Science — building systems at the intersection of large-scale analytics, generative AI, and real-world deployment. My work ranges from analysing 260M+ government transactions to shipping GPU-accelerated RAG pipelines and full-stack analytics platforms. I care about the gap between a Jupyter notebook and a system that actually works in production.
- 🔬 Research: Large-scale data pipelines · synthetic data · retrieval-augmented generation
- 🛠 Engineering: End-to-end ML systems · LLM inference · agentic AI
- 🌏 Domains: FinTech analytics · digital identity systems · maritime intelligence
| 🏅 | Best Outgoing Student — BE CSE-DS, SJEC 2022–2026 |
| 🎖 | Deen Dayal Upadhyaya Fellowship — 10 awarded nationally per year |
| 🥈 | Runner-Up, CIDeCode 2024 — PES University (VarunNetra) |
| 🔟 | Top 10, TonHacker House — India Blockchain Week 2024 (Agriscrow) |
| 📊 | UIDAI Data Hackathon 2026 — 260M+ Aadhaar transaction analysis |
| 🎗 | NCC 'B' & 'C' Certificate — First Rank in unit |
| 📈 | Rank 8, KRLMPCA · KSEEB 10th Board: 96.16% |
DataSynth — Synthetic Tabular Data Generation Platform
Human-in-the-loop EDA platform · 90%+ statistical fidelity across 3 generative model backends
Streamlit-based platform for intelligent data preprocessing, interactive analysis, and synthetic data generation. Ships CTGAN, TVAE, and Gaussian Copula backends with quality validation comparing synthetic vs. original distributions.
Python Streamlit SDV CTGAN TVAE Gaussian Copula Plotly Scikit-learn
Aadhaar Data Insights — National-Scale Identity Analytics
UIDAI Data Hackathon 2026 · 260M+ transactions · 39 states · 6 zones · R² 0.85–0.90
End-to-end analysis of 260M+ Aadhaar transactions across 39 states and 6 geographic zones. Applied K-Means clustering (k=3) to surface enrollment archetypes and built a regression forecasting model achieving R² 0.85–0.90. Delivered 13 interactive Plotly dashboards across temporal trends, MoM/QoQ growth, geographic segmentation, and age-cohort behaviour.
Python Pandas Scikit-learn Plotly K-Means Linear Regression Jupyter
VarunNetra — Maritime Situational Awareness System
🥈 Runner-Up, CIDeCode 2024 — PES University · RAG · FAISS · Real-time Socket.IO
RAG-powered naval intelligence platform for real-time maritime surveillance. Ingests 100+ communication messages, embeds them with Sentence-Transformers into a FAISS vector store, and serves a natural language query interface backed by Gemini. Automated threat detection engine flags unauthorized vessels, distress signals, and restricted-zone infractions in real time.
Python Flask FAISS Sentence-Transformers React Socket.IO Gemini API Leaflet
Feedback Analytics Platform — Full-Stack NLP Feedback Engine
Zero external NLP dependencies · 200+ term lexicon · 5-dimension weighted scoring
Full-stack feedback platform with a custom-built NLP engine. Sentiment scoring uses a 200+ term lexicon (−5 to +5), negation detection, and intensifier amplification. Composite performance score across 5 weighted dimensions: Completion Quality (25%), Rating (25%), Sentiment (20%), Engagement (15%), Response Volume (15%).
FastAPI React 18 MySQL SQLAlchemy JWT Custom NLP Recharts Vite
GPU-Accelerated RAG API Agent — LLM Inference at Scale
Llama 3.3-70B · vLLM · AMD MI300X (192GB HBM3) · Docker + ROCm containerised
High-throughput RAG system running Llama 3.3-70B-Instruct via vLLM on AMD MI300X. Processes PDFs, HTML, and JSON APIs, chunks and embeds into FAISS, and answers questions with context-grounded responses. LangChain orchestration with parallel multilingual query support.
Python vLLM LangChain FAISS FastAPI Docker ROCm AMD MI300X
Agriscrow — Decentralised Agricultural Marketplace
🔟 Top 10, TonHacker House — India Blockchain Week 2024 · TON Blockchain · Smart Contracts
Blockchain-powered agricultural marketplace on the TON network. Smart contract–based escrow with multi-sig requirements, timelock mechanisms, and dispute resolution. Supports four user roles, equipment rental with FCFS scheduling, and native TON cryptocurrency payments.
React 18 TypeScript TON Blockchain Tact Flask SQLite TON Connect Tailwind CSS
Software Development Intern (AI/ML) · CloudxLab · Mar 2025 – Nov 2025 · Remote
Authored six technical deep-dives on applied AI in healthcare published on aiforhealthtech.com, covering synthetic data generation, genomic analysis (DeepVariant), AI-assisted surgical systems, and healthcare data privacy standards.
→ Synthetic Data in Healthcare · DeepVariant: AI with Precision Healthcare · The AI Revolution in Surgery
Languages Python · SQL · JavaScript · TypeScript · LaTeX
ML / AI PyTorch · Scikit-learn · LangChain · FAISS · vLLM · SDV (CTGAN, TVAE, Copulas)
Data Pandas · NumPy · Plotly · Matplotlib · Seaborn
Backend FastAPI · Flask · SQLAlchemy · JWT · MySQL · SQLite
Frontend React 18 · Vite · Tailwind CSS · Recharts
Infra Docker · AMD ROCm · Git
Blockchain TON · Tact · TON Connect
B.Tech — Computer Science & Engineering (Data Science Specialisation)
St. Joseph Engineering College, Mangaluru · 2022 – 2026
CGPA: 9.25 / 10 · Best Outgoing Student, CSE-DS Department



