Work on Intuitive Assistant and GenAI products

Algorithm development for time series change point detection and clusteringCreate and update database in Snowflake using DatabricksData visualization for system performance using Tableau

Built deep learning models to predict optimal read voltages and drive failure rates based on history data using TensorFlow and PyTorch Developed a recursive algorithm to reduce drive read failure trigger rateDeployed the model onto the chip using machine learning framework

Utilised Elastic Search with fast-cosine-similarity plugin to query top-k embeddings for different behaviour networks;Utilised multi-threading, multiple clusters, product quantisation and Faiss to improve search speed;Integrated Elastic Search query module into Airflow Data pipelines;Re-designed textual and numerical features using Spark and determined feature importance using extra tree classifier;Utilised spark SQL to build nodes and edges for Share-IP behaviour graph;Utilised Min-hash to encode raw features and performed connected components and community detection to generate unique user clusters;Visualised unique user clusters using neo4j to demonstrate strong connections based on fuzzy evidence.

Research assistant at Robotics And Rehabilitation (RoAR) Lab directed by Prof. Sunil Agrawal• Participated in designing control algorithm for the mobile companion robot;• Exam the viability of replacing Vicon with IMU using Raspberry Pi to synchronize the signal;• Simulated the arm swinging process in Matlab;• Manufactured parts of the device using laser cutting;• Participated in the experiments designed to validate the effect of stability improvement and analyzed results.

• Utilized TensorFlow to implement IRT model to predict the students’ mastery of knowledge;• Used Hive SQL to determine the statistics about the effectiveness of e-learning recommendation system;• Implemented Trans E algorithm to realize knowledge graph embeddings of large dataset using TensorFlow.