bibhabasumohapatra / bibhabasumohapatra

Hi there ,I'm Bibhabasu

PyTorchian, Computer Vision, NLP, Quantitative Mathematics, Data Science

About me : I overfit xgboost on a daily basis, and train my transformers.

I am a movie person on the weekends and Deep Learner and Machine Learner on the weekdays ,I discuss ml stuff on kaggle ...and try to engage in discussions as much as possible. I try lot of kaggle competitions and learn from kaggle competitions. Currently undergrad at IIIT Bhubaneswar in Computer Science and Engineering.

Focussing on Deep Learning and its application in Computer Vision and Natural Language Processing. And looking for Internships.

• 🔭 I’m currently working on Kaggle Competitions - Sartorius - Cell Instance Segmentation AND Jigsaw Rate Severity of Toxic Comments
• 🌱 I’m currently learning Natural Language processing from CS224N Stanford youtube and Deep Learning - Aaron Courville, Ian Goodfellow, and Yoshua Bengio
• 👯 I’m looking to work in Research Internships - and Data Science Internships
• 🤔 I’m looking for help with Natural Language Processing and MLOps

Kaggle Participation :

• 

Some Notebooks and repos

• Won a bronze medal rank 80th Top 7% (80/1245) in the HuBMAP + HPA - Hacking the Human Body Competition (Image Segmentation) hosted on Kaggle:

  • well managed Github Repo of the Implementation GitHub Repo

  • • Kaggle link for Jupyter Notebook Submission Notebook best Result

• clean Only PyTorch Approach with Huggingface Models - US patent phrase to phrase matching kaggle Competition

  • Training NB - https://www.kaggle.com/code/bibhabasumohapatra/train-us-patent-part-1-simple-pytorch

  • Inference - https://www.kaggle.com/code/bibhabasumohapatra/inference-us-patent-part-1-simple-pytorch

• Top 40% Solution to Pets Pawpularity Kaggle Competition (Image Classification) :

  • Kaggle link for Jupyter Notebook https://www.kaggle.com/bibhabasumohapatra/not-only-the-meta-data-but-ensembled

  • well managed Github Repo of the Implementation https://github.com/bibhabasumohapatra/Pawpularity-Kaggle

• Decline Curve Analysis on a real- life Oil Production Data :

  • Using a pytorch and fitting parameters on a hyperbolic Equation and Exponential Equation and backpropagation and custom loss function . Comparing the resultant graphs with the graphs after applying scipy optimize curve fit. https://github.com/bibhabasumohapatra/Decline-Curve-Analysis

• Ensembles with kfolds some Tabular Playground Series Experiments :

  • Tutorial to Lightgbm www.kaggle.com/bibhabasumohapatra/october-playground-with-lgbclassifier and

  • Tutorial to Xgboost Classifier https://www.kaggle.com/bibhabasumohapatra/xgboost-kfolds-october-playground

Currently Contributing and focussing on PyTorch Ignite and Kaggle Competitions