Hello, I'm Pushpanshu Tripathi

Advanced Institute of Manufacturing with High-Tech Innovations, CCU, Taiwan

Professor Guo-En Chang, NCCU Taiwan

TEEP@AsiaPlus

Modeled the resonance pattern of a gold nanoparticle on a waveguide under different layer width conditions & electric field. Studied the field impact of the nanoparticle filter on a photonic crystal biosensor used for rapid cytokine testing label-free. Developed a full simulation for analysis of localized surface plasmon resonance causing red shift in gold nanoparticle filter.

GEP Worldwide Pvt. Ltd.

Developed time series forecasting models with ARIMA, Holtz Winter for cost analysis in procurement strategy for clients. Implemented an Anomaly detection algorithm with Isolation forest to find anomaly in suppliers data for raw material cost. • Set up functions to find prediction interval and confidence interval for the forecasted value using different models.

Interview Cracker, Blitzrecruit Pvt. Ltd.

Developed Flask APIs for server side implementation of ML models, outputting calculated values of each candidates skills. Implemented an Accent classifier for sales recruitment hirings by training sklearn, MLP, CNN deep learning models. Deployed live users rating prediction system using applicants voice, text input using librosa, gramformer, tensorflow. Used gramformer model to obtain correct grammatical sentences in candidates speech output for rating speaking ability

Battery Cycle Life Prediction

Professor Kanwar Singh Nalwa, IIT Kanpur

Developed a Data-driven approach for prediction of battery life cycles in use of different materials in rechargeable battery. Implemented a non-linear classifier model for prediction of low/high cycle battery with different materials like lithium. Extracted features like charge and discharge capacity for battery cycle life dependence based on different crystal structures.

Speech Emotion Recognition

Implemented CNN, LSTM, MLP models to interpret human emotion trained on speech input from RAVDESS dataset. Developed a speech recorder using pyAudio for deploying the model in real-world applications for interpreting emotions. Achieved 78 % accuracy using CNN model, 73 % accuracy using LSTM model 62% accuracy using MLP model.

Machine Learning and Data Analysis for peak fitting and intelligent prediction

Professor Shikhar Krishn Jha, IIT Kanpur

Training a ML model to study various noise removal and apply it further to X-ray diffraction pattern of same metals to different generated samples automatically.Reducing the time taken for managing datasets obtained from XRD patterns in synchrotron labs by a much larger amount and making the full process automatic.

Autonomous Navigation in Mars Rough Terrain Environments

Simulated 6-wheeled rover and a companion drone to create an autonomous navigation stack in mars terrain environments. Configured GPS sensor, 2D LiDAR and wheel servo motors via the use of Robotics Operating System controllers. Executed A* algorithm for globalized path-planning, used openCV on Depth camera for localized obstacle avoidance

Pet Cat: Quadruped Bot

Collaborated with team to build a bio-mimetic, interactive robotic cat with features for recognition of faces and navigation. Configured the dynamics of the CAD model in simulation for rigorous software testing of controllers via Gazebo software. Implemented obstacle avoidance based on 2D LiDAR and depth camera output, which results in global and local awareness

Canteen-Book App Development

Developed a Canteen-Book App for Android/IOS to be used at canteens/resturants for both keeper and customer side to maintain a transparent transaction of cash between them by keeping a record of all the exchanges on server.