Aman Sah

I

Cracking Binary PUF Hardware Security Using Machine Learning

Indian Institute of Technology Kanpur (IITK)

May 2025 - Current (9 months)

• Research Project based on how the 1980s 2048 randomly connected PUF-CDU Hardware Encryption can be cracked using Linear Regression. • Out of Millions of Combinations our model can crack it within 0.3 seconds using minibatch gradient descent, NAG acceleration, and ISTA to enhance the accuracy of input prediction.

A

Oral Cancer Detection Device

Airshed India

Aug 2024 - Feb 2025 (7 months)

• Muunh Parikshhak is a medical device that can predict oral cancer at an early stage of the diagnosis without undergoing a painful incision or biopsy. • The device utilizes Blue UV rays emitted at a particular wavelength to excite the cancerous cells and capture images. It further utilizes computer vision and several ML - models(Xgboost , RandomForest , TinyVGG , voting classifiers) trained on a huge dataset to predict the stages as Malignant / Pre-Malignant / Normal correctly.

Flight Fare Prediction using Machine Learning

iNeuron.ai

Jun 2023 - Aug 2023 (3 months)

• Conducted Exploratory Data Analysis(EDA), feature extraction on independent variables to enhance data quality and information representation. • Assessed multicollinearity using Variance Inflation Factor (VIF) for the robustness of the predictive models to improve the feature importance. • Achieved an impressive 81.92% accuracy on test data post extensive hyperparameter tuning on Decision Tree,Random Forest Regression, XGBRegression. • Successful model deployment using Flask framework on Google Cloud Platform (GCP).

I

Hexadecimal Image Parity Classification

IIT-KANPUR

Jan 2022 - May 2024 (2 years 5 months)

• Using ResNet-18, a powerful CNN architecture, to tackle the challenge of classifying the parity of hexadecimal numbers in CAPTCHA images. By leveraging ResNet-18's deep hierarchical feature extraction capabilities, achieved 99% accurate predictions through the analysis of rotated, color-varied characters.

About

Detail

Timeline

Résumé