About:
Hi, I am Ash.
I am a machine learning researcher and engineer with over ten years of combined experience in both industry and academia. Holding a PhD in Theoretical and Computational Physics, and a second PhD (defending soon) in Machine Learning. My academic rigor is underscored by 7 peer-reviewed publications. Beyond my academic achievements, I have a profound passion for designing ML systems and have worked on and developed multiple open-source projects. I have also helped design and develop production quality ML solutions at large and small companies, gaining invaluable knowledge during the process.
Recent Work:
Open-Source Project:
Network Embedding Exploration Tool [NEExT]
I have been working on an open-source project called Network Embedding Exploration Tool (NEExT) for the past year. NEExT is a Python project designed for exploring collections of networks, such as proteins or social-graphs. My goal with this project is the following:
- Build an efficient and automated way of computing node and structural features on sub-graphs in the graph-collection.
- Use explainable (human readable) features to compute computationally efficient embeddings (using approximate Wasserstein technique) of sub-graphs.
- Use top-down and bottom-up techniques to explore what low-level features contribute to various graph level embeddings. (such as probing)
This work is still in early stages. We have finalized a paper using this framework, which will be coming out soon. Feel free to `pip install` it and provide some feedback. Thank you.
NEExTRecent Posts:
Generative Model for Graphs
Exploring a generative model for graphs and an overview of GraphRNN.
ReadSelf-Attention Mechanism
Exploring the idea of the Self-Attention Mechanism in Machine Learning.
ReadSimple Text Summarization LLM API Using Flask
Building a simple LLM text summarization API using Flask and Hugging-Face.
ReadSources of Error in Machine Learning
Intuitive overview of various sources of error in machine learning.
ReadRecent Talks:
Unsupervised Framework for Evaluating Structural Node Embeddings of Graphs
Toronto Fields Institute: 18th Workshop on Algorithms and Models for Web Graphs.
Watch