Working to analyze wearable data using Machine Learning techniques

- creating intern training programs- helped create workshops and collaborations with other clubs- Tech Lead for Self-driving Car Project

- Worked on creating a text-based dataset interface using LLMs to generate SQL at a CBInsights top 100 AI Startup- Created an API to generate semantic layers for Salesforce to navigate their customers’ databases using PostgreSQL

Monash Vision and Learning LabAnnotations lead for human pose forecasting in JRDB 2022Corrected 2.4 million bounding box annotations of people in various real-life scenes captured by the Jackrabbot robotWorked with a team of grad students to release the novel dataset for a workshop in the prestigious European Conference on Computer Vision (ECCV)

Developed, designed, and implemented a novel transformer-based method, named SoMoFormer, for multi-person human pose forecasting.Achieved the lowest Euclidean distance between predicted and ground-truth poses.Results published in a paper that is currently published on Arxiv and submitted to the Robotics Automation Letters Journal (RAL)Advisors: Prof. Ehsan Adeli (Stanford University) and Prof. Hamid Rezatofighi (Monash University)

Stanford Vision and Learning Lab (SVL)Developed a text-to-image synthesis model working with Prof. Fei-Fei Li and Dr. Ehsan Adeli Used Vector Quantized Variational Auto-Encoder methods applied to Transformers.Published results in a paper in the Journal of Student Research