┌────────────────────────────────────────────────────────────────────────────┐
│ ██╗ ██╗███╗ ███╗ █████╗ ███╗ ██╗ ██████╗ │
│ ██║ ██║████╗ ████║██╔══██╗████╗ ██║██╔════╝ │
│ ██║ ██║██╔████╔██║███████║██╔██╗ ██║██║ ███╗ │
│ ██║ ██║██║╚██╔╝██║██╔══██║██║╚██╗██║██║ ██║ │
│ ╚██████╔╝██║ ╚═╝ ██║██║ ██║██║ ╚████║╚██████╔╝ │
│ ╚═════╝ ╚═╝ ╚═╝╚═╝ ╚═╝╚═╝ ╚═══╝ ╚═════╝ │
├────────────────────────────────────────────────────────────────────────────┤
│ [SYS] mounting /dev/consciousness ................................. OK │
│ [GPU] cuda cores detected ......................................... OK │
│ [NET] uplink → IIT Mandi // Affective Computing Lab ............... OK │
│ [SIG] fourier filter on EEG stream ................... SNR +8~10% OK │
│ [OPT] cnn.prune() ................................. 1.2M → 156K OK │
│ [WRN] sleep daemon not responding .......................... IGNORING │
│ [OK ] operator ready. │
└────────────────────────────────────────────────────────────────────────────┘
operator:
handle: UMANG-SH941
designation: Umang Shukla
class: Deep Learning Researcher ✦ CS (AI & ML) Undergrad
home_node: Republic Of India
education:
institute: Jaypee University of Engineering & Technology, Guna
program: B.Tech CSE — AI & ML Specialization
window: Aug 2024 → Jul 2028
coursework: [DSA, Linear Algebra, Probability Theory, Multivariate Calculus,
Computer Vision, Deep Learning, Design & Analysis of Algorithms]
current_directives:
- researching: "affective computing + spatiotemporal architectures for
emotion recognition in low-resource settings"
- forecasting: "energetic particle flux around GEO — single-event-upset risk
for ISRO satellite electronics"
- shipping: "EvoSnap — AR sustainability scanner, signed APK in the wild"
query_me_about: [PyTorch, Signal Processing, Model Compression,
Bayesian Stats, Computer Vision, Space Weather]
motto: "When you have no legacy to inherit, you create one."
| STACK LAYER | MODULES |
|---|---|
| Languages | Python (proficient) · C · C++ · SQL (PostgreSQL, SQLite) |
| Statistics & Optimization | Bayesian statistics · Probability theory · Convex optimization · Model fitting · Multivariate calculus · Linear algebra (eigendecomposition, rank-nullity) |
| Machine & Deep Learning | PyTorch · Scikit-Learn · CNN · RNN · LSTM · Transformers · Siamese Networks · U-Net · Transfer learning · Ensemble methods |
| Data & Signal Processing | NumPy · Pandas · OpenCV · SciPy · Fourier filtering · Epoching |
| MLOps & Infra | AWS (S3, EC2, SageMaker) · Docker · Kubernetes · Linux · Git/GitHub · FastAPI · REST |
●━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━►
│
├─ AUG 2024 ▸ JUET Guna — B.Tech CSE (AI & ML) begins
│
├─ OCT 2025 ▸ Hacktoberfest — 15+ PRs merged across repos
│
├─ OCT 2025 ▸ IIT MANDI — Student Intern, Deep Learning Research
│ ├ Fourier-filtering + spatial-normalization for high-noise EEG
│ ├ SNR & classification accuracy ▲ 8–10%
│ ├ Parameter-efficient CNN: 1.2M → 156K params (−87%)
│ └ Multi-threaded PyTorch loaders: pipeline latency ▼ 35%
│
├─ 2025 ▸ ICVGIP 2025 — Organizing Team (submissions + speaker logistics)
│
├─ MAY 2026 ▸ IIT MANDI — Research Intern, Affective Computing & CV
│ ├ Convolutional / transformer / spatiotemporal architectures
│ └ Reproducible benchmarking + rigorous validation protocols
│
└─ NOW ▸ ISRO GEO radiation forecasting · IEEE SAC Vice Chair
React Native Expo TypeScript FastAPI ONNX MobileNetV3
AR mobile app with a FastAPI/AI backend that pairs a vision-language model with a self-trained MobileNetV3 classifier (TrashNet, 89% accuracy) to identify an object and simulate its environmental decay over time. Multi-stage sustainability reasoning — carbon footprint, recyclability, decomposition timeline — with automatic fallback across three image-generation providers, offline-first persistence, and a signed Android APK release.
Python Time-Series Forecasting ML Space Weather
Time-series forecasting of energetic particle flux to characterize the radiation environment around geostationary orbit, supporting risk assessment for onboard satellite electronics against radiation-induced single-event upsets.
| ▸ | PROJECT | RESULT | STATUS |
|---|---|---|---|
| ☄️ | EvoSnap | 89% classifier accuracy · signed APK shipped |  |
| ☢️ | ISRO GEO Radiation Forecast | SEU risk characterization |  |
| 🧠 | EEG Pipeline @ IIT Mandi | SNR & accuracy ▲8–10% · latency ▼35% |  |
| 🗜️ | Parameter-Efficient CNN | 1.2M → 156K params (−87%) | |
| 🎭 | Affective Computing Benchmarks | reproducible validation protocols |  |
| CMD | ACHIEVEMENT |
|---|---|
| 🎖️ | IEEE Student Activities Committee — Vice Chair. Coordinating 35+ colleges across the region. |
| ⚡ | IEEE Xtreme Competitive Programming — Top 8% globally among thousands of international teams. |
| 📄 | ICVGIP 2025 — Organizing Team. Paper submissions + speaker logistics for a premier Indian CV conference. |
| 🛸 | IETE Students Forum — Technical Lead. Guided 50+ students in hardware-software interfacing and algorithmic design for drone development. |
| 🌍 | Hacktoberfest 2025. 15+ PRs — preprocessing utilities, CI pipelines, code-review workflows. |
╭──────────────────────────────────────────────────────────────╮
│ "When you have no legacy to inherit, you create one." │
╰──────────────────────────────────────────────────────────────╯