MADHUVANTHI VENKATESH

Results-driven data engineer and BI expert with 6 years of experience in building scalable data systems, ETL pipelines, and cloud-native solutions. Proven track record in delivering impactful data products using AWS, Spark, and modern BI tools. Skilled in developing end-to-end data pipelines, machine learning workflows, and intuitive dashboards with Power BI, Tableau and QuickSight. Recognized for fast onboarding, strong collaboration, and a proactive, learning-oriented mindset. Passionate about data science, geospatial analytics, and turning complex data into actionable insights.

• Digitize IT- Philipps University of Marburg and Technical University of Munich [Master Thesis]
• April 2020 – December 2021
• Works at Philipps University of Marburg:
• Developed a software toolbox that extracts maps from the textbooks, identifies the elements within textbooks, georeferences them using GDAL function, creates masks and georeferences the masks that contain the occurrence points, and finally prints the coordinates of the maps as well as the centroids of the occurrence points.
• Used the GCP points from the output of QGIS to do batch georeferencing on the input files.
• Works at Technical University of Munich:
• As a part of my master thesis, tutored the theory and tutorial of Digitize IT to students and made them process the software toolbox for 14 different input textbooks.
• GitHub: https://github.com/environmentalinformatics-marburg/distribution_digitizer_students

• National Atmospheric Research Laboratory (A part of Indian Space Research Organization (ISRO)), Gadanki, India [Bachelor Thesis]
• Jan 2017 – March 2017
• Under the supervision of the scientist, Dr. K. Ragunath.
• Works:
• The internship involved studying of 300 MHz High-frequency Amplifier Discriminator and redesigning the circuit for discriminating frequencies more than 300 MHz. This was used in LIDAR Communication.

BERT-AI-Text-Classifier (Winner - DataNova 2025)

This project evaluates four encoder-based NLP architectures — BERT, RoBERTa, DistilBERT, and TinyBERT — trained on a common dataset. It compares their classification accuracy, efficiency, and inference speed to analyze performance–size trade-offs.

GitHub: https://github.com/VenkateshMadhuvanthi/BERT-AI-Text-Classifier/

Found it using Gemma (Runner Up – LablabAI Gemma Challenge 2024)

This project introduces an AI-powered assistive application designed to support elderly individuals with memory challenges. Using Gemma models and video data, the system detects and tracks everyday items, storing frame embeddings and metadata in ChromaDB for efficient query-based retrieval—allowing users to ask questions like “Where did I last see my pen?” Through a simple and intuitive interface, the application helps users locate items independently, reducing frustration, supporting caregivers, and providing peace of mind for families, especially those living abroad.

Link: https://lablab.ai/event/gemma-2-ai-challenge/trendseekers/found-it

NN-Classification on Parkinson’s Disease - Philipps University of Marburg, Germany

A project that involves Keras and SDCM to classify the dataset of the history of patients who had Parkinson’s Disease and predict the characteristics from the records of the new patients., 

GitHub: https://github.com/VenkateshMadhuvanthi/NN-Classification-on-Parkinson-s-Disease

A High Accuracy Framework in Automatic Gate Control and Monitoring - Anna University, India
Project to address the drawbacks observed in the previous research papers and a new model was developed for the complete and foolproof security of railway gating system to avert accidents.

Publication: “A High Accuracy Framework in Automatic Gate Control and Monitoring”, at the International Conference on Frontiers in Engineering, Applied Sciences and Technology (FEAST 17), NIT Trichy, Trichy, India, Page 176-180, ISBN 978-81-908388-8-7.

AWS Data Engineer - Associate

TinyML and Efficient Deep Learning Computing

Tableau

Grafana

Docker