Electronics Engineering Undergrad ('22–'26) | ML Optimization & Hardware Acceleration
I'm obsessed with making ML efficient. Not just accurate—efficient. Quantization, pruning, distillation, edge inference. ECE background means I think in hardware constraints, not just model metrics.
Current Interests: ML optimization, hardware acceleration, edge inference.
| Category | Tools & Technologies |
|---|---|
| AI/ML |     |
| Systems/Backend |    |
| Cloud & DevOps |     |
| Frontend/Data |    |
Sign Language Recognition — Optimization flagship
- 99.88% validation accuracy on ASL fingerspelling
- MediaPipe → PyTorch MLP
- Roadmap: quantization, pruning, distillation, edge deployment
- 🔗 Code
Go Web Crawler — Systems thinking
- 10k+ URLs in <5 seconds
- Goroutines + channels. Because speed is an optimization problem.
- 🔗 Code
Contract Intelligence Parser — Inference at scale
Unsupervised Fraud Detection — ML fundamentals
- 0.93 ROC-AUC on 280k transactions (GMM)
- 📓 Notebook
I approach ML optimization from first principles—paper-first, implement-from-scratch. Currently working through deep learning fundamentals: MLP → LeNet → VGG → ResNet → RNN/LSTM → Transformers. Understanding the architecture, not cargo-culting the code. Learning Japanese (JLPT N4/N5 target: December 2026) because I'm serious about working in Japan's optimization ecosystem.
Off-duty? Competitive programming, systems design problems, and reverse-engineering how things actually work.
Open to OSC in quantization/optimization toolchains. Let's collaborate on making ML faster and leaner.
