• Build long-short trading strategies and back test them on previous data using mathematical models• Made use of fast-expression language to build 60+ alphas with sharpe ranging from 1.5 to 4.5 in USA, China, Asia, and various other regions

• Successfully scaled the society to be one of the largest in the university within 6 months of inception.• Took sessions on AI/ML, Web Development and App Develoment with over 100+ attendees.• Hosted events in collaboration with H2O.ai, Weights and Biases and many other such big tech companies

• Worked on Research for toxic comments and developed the SOTA algorithm to find the toxic comments.•Worked on synthesizing Arabic and Bengali handwriting using LSTMs

• Developed a novel LLM training pipeline, encompassing both custom Python and C++ code for efficientalgorithms and data structures• Developed an end-to-end prompt library, significantly reducing inefficiencies by streamlining prompt management

• Utilized Celery’s asynchronous task execution to improve pipeline responsiveness and overall throughput by 3 times.• Developed the backend of the sales call research automation tool using Django

• Developed a workflow to filter out IMS notices based on course and semester, allowing for more targeted and relevant communication with students.• Took a speaker session on hands-on deep learning with over 100+ attendees

Successfully planned and executed a machine learning competition with over 120 participants.Developed a unique dataset with custom fields for the competition

• NFL Data Analytics on 40GB video dataset using Temporal Action Recognition.• Created a dashboard making use of explainable AI using SHAP for detail sport analytics of players and teams