B.Tech Mathematics & Computing · IIT Mandi (2023 – 2027)
Physics-informed neural networks, quantitative finance and full-stack engineering.
Website · LinkedIn · rajmahimaurya@gmail.com · Résumé
I build physics-informed models — and the software around them. Most of my time goes into PINNs for fluid dynamics: neural networks that learn the solution of a differential equation from the equation itself, where a well-chosen change of variables or ansatz does more than another layer. Alongside that I study stochastic models of markets and keep shipping full-stack products.
SAGE-PINN: A Singularity-free Axisymmetric Geometry-Encoded Physics-Informed Neural Network for Axisymmetric Multiphysics Flow
Raj Maurya, Joyprakash Akhuli, Doyel Pandey — arXiv:2610.10449 (2026), submitted to Journal of Computational Physics
Removes the 1/r axis singularity of cylindrical PINNs by construction (s = r², a modified stream function, exact boundary ansätze). Trained on physics alone, one setup solves 40 nanofluid stenosis cases (Re 0.46–128, Pe 10.5–2330) to within 1.6% axial-velocity error against finite-element references.
Also: a follow-up on electroosmotic flow in nanochannels (in preparation), replications of PINN blood-flow and wavy-channel studies, and Neuro-Trade — Fokker–Planck PINNs for Indian equities.
| Project | What it is | |
|---|---|---|
| JourneySync | Real-time cross-channel identity resolution — AMEX CodeStreet 2026 finalist (top 6) | Python · PostgreSQL · Docker |
| ByteChat | Anonymous, encrypted chat for the IIT Mandi community — bytechat.in | Next.js · Supabase · AWS |
| C-TAEA | From-scratch constrained multi-objective evolutionary algorithm (IEEE TEVC) | Python · pymoo |
| Deepfake Detection | Fully unsupervised detection via enhanced contrastive learning | PyTorch |
| GenWeb | AI website generator — describe it, agents build and deploy it | Next.js · Supabase |
| NetSentinel | Real-time network traffic monitoring and threat detection | React · FastAPI · Scapy |
| PDF-GPT | Private, local-first chat over your PDFs with Ollama | Python · Ollama |
| Check-It | In-browser real-time object detection, no server | TensorFlow.js |
All ~50 projects, sorted and searchable: rajm012.vercel.app/projects
- Research, physics-informed ML for fluid dynamics — IIT Mandi, with Dr. Doyel Pandey · 2026 – now
- Software Development Intern — Pinch · May – Jul 2026
- Teaching Assistant, Software Development with AI — Himshikhar, IIT Mandi · May – Jul 2026
- Research Intern, computational complexity — IIT Mandi · Dec 2024 – Jan 2025
- Core Member — Research Society, IIT Mandi · 2024 – now
Languages Python, JavaScript/TypeScript, C++, Bash, SQL ML & scientific PyTorch, TensorFlow, scikit-learn, NumPy/SciPy, COMSOL Multiphysics, MATLAB, Qiskit Modelling PINNs, ODEs/PDEs, SDEs, Fokker–Planck, optimisation Web & infra React, Next.js, React Native, Node/Express, GraphQL, FastAPI, Flask, PostgreSQL, Supabase, Docker