Paritosh Sarangi

Experience

• 

  Chi SquareX

  Feb 2022 - Apr 2022

  Machine Learning Developer

  Image Segmentation and Transfer Learning with Python•Used Breast Cancer data from Kaggle; the source cited was “Al-Dhabyani W, Gomaa M, Khaled H, Fahmy A, Dataset of breast ultrasound images. Data in Brief. 2020 Feb;28:104863. DOI: 10.1016/j.dib.2019.104863”•Did Data Augmentation using the Albumentations library, thus increasing dataset size from 780 images and their corresponding masks to 10,000, for improving model adaptability.•Created an Image Segmentation model using UNet architecture to generate masks for the ultrasound images, that is, differentiating the tumor from the other lesions and masses of the ultrasound. Achieved F1 Score of 0.9912.•Did Transfer Learning with ResNet50 as the base model to give a diagnosis on the nature of the tumor based on the mask generated by the Segmentation Model. Achieved accuracy of 0.9773 and F1 Score of 0.9754. Show less

• 

  Indian Institute of Technology, Kharagpur

  May 2022 - Jan 2023

  Research Intern

  Published in Physics of Fluids(DOI: 10.1063/5.0208554)Prediction of Fluid Flow Characteristics within rough surface microchannel Objective: Building a Machine Learning Model to predict pressure difference across a microchannel during fluid flow based on the surface roughness characteristics, deduce how the pressure difference varies concerning the parameters, and use it to deduce other fluid flow parameters.•Gathered data by creating simulations on COMSOL MultiPhysics software for getting pressure difference values across microchannels with varying parameters(surface roughness, inlet velocity, pipe width, slip length, Reynold's Number).•Used different Machine Learning and Deep Learning Models to find the best-fit model for the data. The highest R2 score was achieved using ANN(0.9963).•Used the model to predict and plot the output parameters against input parameters in different situations, and deduced empirical correlations between the different output parameters(vorticity, shear rate, Pressure Difference).• Compared the results obtained from machine learning with results obtained using numerical methods on COMSOL and past experimental observation data to verify the findings and deductions from the ML results. Show less

• 

  HDFC Bank

  May 2023 - Jul 2023

  Data Science Intern

  Worked on building modules to help compute Feature Importance values for Gradient Boosting models on Out of Time, Out of Sample and Post Development datasets.Helped improve reliability of feature stability evaluation across time for XGBoost and LightGBM models.

• 

  HDFC Bank Limited

  Jul 2024 - now

  Management Associate