I work on data-heavy projects in Python, focusing on preprocessing, structuring datasets, and organizing experiments into clean, reproducible code.

Below are some examples of my work:

• EDA & ECG ML pipeline [Tilburg Uni] [Ongoing]
  Developing structured preprocessing and machine learning/analysis pipelines for physiological time-series data (skin conductance and ecg), including cleaning, validation, and feature preparation. Currently looking into feature importance using recursive feature elimination. https://gitlab.uvt.nl/tsb-rst/eda_ecg_preprocessing

• EEG / Brain-Computer Interface (Motor Imagery) [Independent] [Ongoing]: Reproducing the motor imagery BCI pipeline developed by Tibrewal et al. 2022 from scratch with modifications. Preprocessing, epoching, artifact rejection (ICA), and using models like CSP and LDA for classification. Repository link here.

• ECG Arrhythmia Classification [Independent] [Finished]
  Built an end-to-end ML pipeline for classifying clinical ECG data, including preprocessing, feature extraction, model training, and evaluation. Read the indepth article about the project and its limitations here.

• EEG / Brain-Computer Interface (code modulated VEP) [AI Dept, Radboud Uni][DataDrivenNeurotechLab] [Finished]
  Developed and evaluated signal-processing + ML methods (CCA, LDA) for classification in a BCI experiment.

• Universal Text Conversion Tool
  Built a Python tool to automate large-scale document parsing and conversion into structured, searchable PDFs across nested directories.
  https://gitlab.uvt.nl/tsb-rst/universal_text_conversion

• Unstructured Text Parsing & Extraction
  Developed an application to collect, organize, and extract structured information from messy and scanned documents using API integrations.
  https://gitlab.uvt.nl/tsb-rst/os_companies_house_api_integration

• Advanced SQL (DataLemur) – practicing industry-style query problems

📫 LinkedIn: https://www.linkedin.com/in/shekharnarayanan

Python3

Matlab

R

SQL

Pytorch

Numpy

Pandas

Sklearn

Matplotlib

Keras

Conda

Jupyter

AWS