Mukul Lambat

• Developed a python-based machine learning pipeline for bat species classification. • Created synthetic bat calls by introducing thermal, stationary and non-stationary wind noise using NumPy & Librosa python library. • Achieved 96% accuracy by combining Random Forest with Mini-Rocket feature transformations.

• Developed a comprehensive data generation pipeline to produce large-scale, topology-optimized 3D structures using python. • Ensured dataset integrity by flagging invalid 3D structures failing compliance and volume fraction criteria, followed by visual checks. • Generated 4000 3D structures and accelerated the synthetic generation by 50x using python parallel processing.

• Developed a python based regression pipeline to predict force from multi-sensor data, applying feature engineering and EDA. • Performed data cleaning, noise reduction, and feature engineering to capture correlations between sensor data for better predictions. • Trained and evaluated 1D CNN, LGBM Regressor, and Stochastic Gradient Descent models achieving 5% Mean Absolute Percentage Error (MAPE) with the 1D CNN.

• Implemented strain gauge sensor data analysis using Python for load conditions and load steps classification. • Conducted data preprocessing, outlier removal and visualization to optimize data quality and enhance the effective presentation of findings. • Achieved 98% accuracy rate in machine learning model performance using Support Vector Machine (SVM).

Artificial Intelligence(AI), Machine Learning(ML), Deep Learning, Digital Signal Processing