I build systems where the hard part is the decision, not the demo — collision-avoidance planning for satellites, self-healing supply chains, transactional shell execution. Third-year Computer Engineering undergrad at VIT Pune, currently a software development intern on the LearnGeeta platform. I care about work that holds up under measurement: benchmarks against a real baseline, tests that mean something, and reproducibility scripts so the numbers survive a second run.

Concepts  Data Structures & Algorithms  OOP  DBMS  REST APIs  Machine Learning  Reinforcement Learning  Quantum Machine Learning

Python Orbital Mechanics React

Screening and manoeuvre-planning system that ranks conjunction risk between satellites and recommends avoidance burns. Across a replay of 8,892 conjunction events it missed zero high-risk events against 18 for the baseline, while using 6.2% fewer manoeuvres and 13.8% less delta-v. Backed by 167 tests and a 78-check reproducibility script.

repository  ·  live demo

AI Optimisation Web

Detects supply-chain disruptions, traces them to a root cause, and plans recovery actions — then feeds outcomes back into the network model so repeat disruptions cause less damage. Ships with a working in-browser recovery solver. First Runner-Up at Hack Genesis, built in a 5-member team.

repository

Linux overlayfs Merkle Hashing

Runs shell commands inside an overlayfs sandbox so any command can be committed or rolled back as a single transaction, and verifies filesystem state with Merkle hashes so the safety guarantee holds regardless of which model or user issued the command. Submitted to the C-DAC Software Samprabhuta Mission as an 18,000-line codebase with a 390-test suite.

demo

Python Quantum Machine Learning TypeScript

Classifies BRCA1 and BRCA2 genetic variants as pathogenic or benign using a hybrid classical-quantum model. Paper in preparation for IEEE PuneCon 2026, Life Sciences track.

repository

Python Computer Vision Remote Sensing

Sub-pixel multi-sensor registration of Chandrayaan-2 imagery against lunar reference maps. Built for Smart India Hackathon 2026, problem statement 26166 (ISRO).

repository

JavaScript Zoho Catalyst AI

AI-powered crime intelligence command center built for the Karnataka State Police Datathon 2026.

repository

_{Also in the cabinet: pariksha (steganalysis of model weights + signed ML-BOM) · netra (on-device status-LED and error decoding) · disaster-response-memory-system (multimodal retrieval with Qdrant)}

First Runner-Up — Hack Genesis, Hackers Occupied Pune (2026). ₹5,000 prize for the SupplyChain AI Control Tower, built in a 5-member team.

Winner — Internal Smart India Hackathon, VIT Pune (2026).

_{Beyond the code, former roles — Creatives Director at Vishwaconclave · Executive Committee Member, IEEE Student Branch, VIT Pune.}

_{Certifications — The Complete Full-Stack Web Development Bootcamp (Dr. Angela Yu) · Supervised Machine Learning: Regression and Classification (DeepLearning.AI) · Advanced Learning Algorithms (DeepLearning.AI)}

_{Generated from the GitHub API by scripts/gen_stats.py and refreshed daily by a workflow, with the DSA panel read live from LeetCode — no third-party stats service, so nothing here breaks when someone else's rate limit does.}

Open to software engineering internships.  Fastest way to reach me is email.