Offered to work full-time starting in August 2020 after successfully working together on Master thesis
Implemented state of the art deep learning models (Quicknat, U-Nets) for liver and spleen segmentation using
nnU-net, Octave-Conv, SWA, SWAG, Thick-slices, Multi-phases
Developed data pre-processing pipelines to prepare 23.000+ MRI scans for input into deep learning segmentation
Built web application using Python/Flask/Docker to visualise MRI scans and its segmentations/heatmaps
predicted by machine learning models; deployed using AWS EC2/S3/Lambda
Machine Learning Research (intern) โ Disney Research Lab (Zurich) -- Nov 2018 - Feb 2019
Tech stack:Python, Keras, Pandas/Matplotlib/Numpy
Collected, cleaned and prepared datasets of facial expressions (e.g. cry, yawn) to be used to classify emotional
reactions of movie viewers and measure which pieces of a movie are engaging
Developed deep learning CNN models to classify 8 basic facial expressions from Cohn-Kanade dataset
Data Engineer (working student) โ Quant-IP (Munich) -- Mar 2018 - Oct 2018
quant-ip.com is a data provider serving large financial institutions
Indian IT services and consulting company, largest company by market capitalization in India
Tech stack:JavaScript/TypeScript, Angular, Node, MongoDB, HTML, CSS, PHP, MySQL
Awarded TCS GEMS award for excellent coding skills among 500 co-workers at the Qualcomm account
Managed and improved data consistency of 200.000+ records across 500 MySQL tables for Qualcomm application
Developed user & administration interface and CRON tasks for Qualcomm learning management web application
Built hybrid app with Angular used by employees of Qualcomm and TCS
๐ EDUCATIONAL
M.Sc. Informatics โ TU Munich (Munich) -- 2017 - 2020
Master Thesis which was accepted at MLMI2020 - https://arxiv.org/pdf/2008.12680.pdf (machine learning
conference on medical imaging): Analysed deep learning model accuracy using PyTorch/scikit-learn/NumPy;
compared 4 state of the art Bayesian neural networks for image segmentation and developed improvements
i-Graph project: Wrote web application in TypeScript using RASA-NLU to translate language into SQL queries
Expression prediction project: Detected emotions on image and video data using PyTorch and OpenCV
Allianz HackaTUM-2019: Won 1st prize out of 37 teams for detecting cracks on metal using image analysis
Bachelor Electronics and Communications โ Gandhi Institute (Gunupur, India) -- 2009 - 2013
Above-average student (top 15%, GPA 8.3); wrote bachelor thesis detecting cracks in railway tracks based on data