Experience
WebOps Secretary
Projects
Power Load Forecasting System
Time-series load classification on 35K+ records at 430 RPS — LightGBM with cyclical feature engineering and GCP deployment.
Genrefy — Audio Classifier
EfficientNet-V2 trained on log-mel spectrograms for 10-genre music classification under distribution shift (clean stems → noisy mashups) — 0.930 on Kaggle private LB.
DeepWaste — CNN vs Fine-tuning Benchmark
Deployment benchmark across 5 inference formats — TensorRT FP16 achieved 6,905 img/s with 98.41% accuracy on garbage classification.
Sales Forecasting — Kaggle Rank 133
Multiplicative forecasting model with Fourier analysis and GDP indicators — 12.7% MAPE across 6 countries, 10 years of data.
Customer Profiling
K-Means & DBSCAN clustering algorithms for customer segmentation analysis with interactive profiling dashboard.
CrossEmotion — Multimodal Emotion Recognition
Fusion of video (R3D-18), text (BERT), and audio (Conformer) with cross-modal attention — 67.8% accuracy on MELD dataset.
Logic-Guided Diagnosis — Neuro-Symbolic AI
Residual MLP trained with differentiable rule-agreement loss — 88.32% accuracy, 0.9516 ROC-AUC on heart disease prediction.
Technical Skills
Languages
- Python
- TypeScript
- JavaScript
- SQL
- Bash
Machine Learning
- PyTorch
- PyTorch Lightning
- Scikit-learn
- LightGBM
- ONNX
- TensorRT
- W&B
MLOps & Deployment
- Docker
- CI/CD (GitHub Actions)
- GCP
- AWS SageMaker
- FastAPI
Databases
- SQL
- DynamoDB
- Neo4j
- Qdrant
Concepts
- Model Optimization
- Time-Series Forecasting
- Computer Vision
- Multimodal Learning
- Neuro-Symbolic AI