ML
Machine learning research, model design, and applied AI
Machine Learning Engineer @ NeuroTech @ Berkeley
Jan 2026 — Present- Achieved 5° RMSE on supervised regression predicting gait angle from 16-channel EEG data for BCI-driven exoskeleton control
- Designed a CNN-LSTM architecture — CNN extracts spatial-spectral features, LSTM performs gait angle regression
- Preprocessed raw EEG with artifact removal, bandpass filtering, per-channel normalization, and overlapping sliding windows
PythonPyTorchCNN-LSTMEEG/BCI
Data Scientist @ Synopsys (Contract)
Aug 2025 — Dec 2025- Trained and benchmarked regression, random forest, gradient boosting, and GNN models to predict chip power consumption
- Evaluated models on MAE, RMSE, and MAPE metrics across ~50–100 chip designs
- Used SHAP analysis to identify dominant features driving power consumption predictions
PythonPyTorchTorchGeometricSHAP
Projects
NeuroTech EEG Gait Decoder
PyTorch pipeline predicting gait angle from 16-channel EEG for BCI-driven exoskeleton control. CNN-LSTM architecture reaching 5° RMSE on supervised regression.
PythonPyTorchBCI
Disaster Response NLP
Sentiment analysis on ~45K disaster tweets across 14 events using RoBERTa and VADER. Batch processing cut inference time roughly in half.
PythonRoBERTaNLP
Disinformation Detection System
Classification system using Logistic Regression with GloVe embeddings and Bag-of-Words features. 91% accuracy and 87% F1-score, iterated via confusion-matrix analysis.
Pythonscikit-learnGloVe