I studied model compression for large natural language models based on transformers. We developed a new method for reducing model size up to 70% while still maintaining same performance and in some cases even improving the performance. Read our work here.

Developed a machine learning model in PyTorch and Python for generating graph embeddings for faster link prediction in graphs. The model is targeted for performing recommendations in an environment that is constrained by data and computational capacity. Project involved working on Node Embedding methods such as node2vec, DeepWalk and Knowledge Graph Embedding methods like TransE, TransH, TransD, SimplE, DistMult & RESCAL. Read our work here.

I worked on an ERP product. My development stack included C#, C++, VB.NET and SQL. I was responsible for an entire business module to remove design flaws and rebuild the module using object-oriented programming practices. We leveraged asynchronous programming to increase the performance and efficiency of the product. Worked on various uses cases, product enhancements, feature development using C# and MySQL and contributed to many successful releases. I also mentored junior developers for programming in C# and C++.

Responsible for fully managing a combined graduate and undergraduate course of ~70 students, including delivering the course material to students, coordinating the grading of assignments & tests with teaching assistants. Collaborated with 5 industry partners for students to work on real world computer vision projects.

Responsible for teaching PyTorch and developing challenge assignments for the course.

Developing a bundle of machine learning programming assignments that can be easily distributed to students and can also be autograded. Engaging students in discussions and helping them understand the subject.

Worked on implementation of Autolab - A course management platform for autograding assignments. I used Docker, Python and Shell Scripting for writing autograders for computer vision problems such as object classification, detection, and autonomous driving.

Educated and Assisted graduate students on Principal Component Analysis, Linear Discriminant Analysis, Logistic Regression, Linear Regression, Neural Networks and Clustering.