M.Sc. Data Science · Deep Learning · Biomedical ML
🌍 Available to work from anywhere
I'm an M.Sc. Data Science candidate at Manipal Academy of Higher Education, focused on deep learning for biological and physiological data. I build GNN-based molecular interaction models, CNN-Transformer hybrids for temporal signal decomposition, and biomedical ML systems deployed in production.
Previously Deep Learning Engineer @ Temple (Eternal), where I worked on sleep onset/offset detection from wearable biosignals (EEG, accelerometer, PPG). IEEE published researcher — always open to roles and collaborations at the intersection of ML, signal processing, and healthcare.
Mar 2025 – Jul 2025 · Gurugram, India
- Developed deep learning models for sleep onset/offset detection using multi-channel physiological time-series
- Designed and trained CNN-LSTM hybrids for local patterns and long-range temporal dependencies
- Built preprocessing pipelines with artifact rejection, bandpass filtering, and VMD-based signal decomposition
- Benchmarked threshold baselines, XGBoost, LSTM, and Transformer architectures for production wearables
| Project | Description | Link |
|---|---|---|
| GNN-DDI Predictor | Graph autoencoder for drug–drug interaction prediction via link prediction on molecular graphs | Live Demo |
| Medicine Recommendation | TF-IDF + cosine similarity system over clinical review data, deployed on Streamlit | Live Demo |
| VMD-CNN-Transformer | Variational Mode Decomposition + hybrid deep learning for non-stationary time-series forecasting | Research |
- IEEE CIEES 2025 — Published paper on benchmarking ML/NLP techniques for public health discourse analysis
Read on IEEE Xplore - Novo Nordisk GBS Hackathon 2025 — Team Lead; delivered a working ML prototype under tight deadlines
Core: Python · SQL · PyTorch · PyTorch Geometric · CNN · LSTM · Transformer · GNN
Domains: Biosignal processing · VMD · RDKit · DrugBank · NLP · Statistical modeling
Tools: Streamlit · Hugging Face · XGBoost · SHAP · scikit-learn
Let's connect — Portfolio · LinkedIn · amantripathi3273@gmail.com