Machine Learning Intern

Duration:

- Total: 3 Months (Unpaid)

- A certificate will be provided after internship completion.

- Work Nature: Remote Mode

Project Overview

- Goal:

Evaluate photoplethysmography (PPG) for cuffless blood pressure (BP) tracking.

- Target:

Early preeclampsia risk detection in pregnant women.

- Current Focus:

Algorithmic limits and physiological validation in healthy adults.

Key Responsibilities

- Data Pipelines:

Clean and filter public physiological waveform datasets (UCI, MIMIC-III, VitalDB).

- Feature Extraction:

Derive time-domain, morphological, and derivative features from PPG signals.

- Model Training:

Develop tree-based models (XGBoost/Random Forest) and easy 1D CNNs.

- Rigorous Validation:

Implement strict Leave-One-Subject-Out (LOSO) cross-validation.





- Robustness Analysis:

Evaluate model performance against acute physiological changes like vasoconstriction.

Required Skills

- Core Tech:

Python, Scikit-Learn, SciPy, and XGBoost/LightGBM.

- Deep Learning:

1D CNN construction in PyTorch or TensorFlow.

- Signal Processing:

Biosignal libraries (

NeuroKit2

,

pyPPG

,

HeartPy

, or

WFDB) (Optional)

- Validation Rigor:

Subject-wise cross-validation and time-series data leakage prevention.

- Domain Mindset:

Physiology-first approach prioritizing vascular physics over black-box fitting.

📌 Machine Learning Intern (Tamil Nadu)
🏢 Hrzworkz Private Limited(HERTZWORKZ)
📍 Tamil Nadu