Harish Kashyap

Harish Kashyap is a machine learning researcher with extensive experience in AI. He has worked in several AI-related organizations, such as Amazon Robotics, and in research labs such as MERL. He holds several patents and has authored numerous publications, and he is an AI subject matter expert at MIT Horizon. Kashyap has authored the AI curriculum for SUNY, Buffalo. He received an MS in electrical engineering from Northeastern University.

Research Interests

• Curriculum and Instruction
• Data Sciences
• Technology and Education

Honors & Awards

• AI Subject Matter Expert, MIT Horizon (2023)
• Best Presentation Award, International Conference on Machine Learning Applications (2017)
• Travel Grant, Pacific Institute of Mathematical Sciences, PIMS, Mathematics Industrial Problem Solving Workshop (2008)
• Graduate Research Award (Research Assistantship), Northeastern University (2005)

Current Research and Scholarly Interests

My research focuses on the development of machine learning and deep learning algorithms for a variety of complex real-world problems, with particular emphasis on signal & image processing, generative AI, and probabilistic methods. My goal is to build scalable, robust, and effective models that can bring about significant advancements in fields such as robotics, perception, and automation.

One of my main areas of interest is the application of machine learning to robotics. During my tenure at Amazon Robotics, I developed deep learning models to significantly improve the efficiency and safety of robotic operations. For instance, my work on the Robin project allowed for the detection of successful picks by the robotic arm through the processing of sensor signals, while my AR-ID project automated the detection of barcodes on packages using computer vision-based methods. Both these projects have patents pending, underlining their innovation and potential impact.

Another area where I've dedicated a significant amount of my research is in the realm of signal and image processing. My work at Voyagenius Labs LLP and as an instructor at Stanford Continuing Studies further underlines my interest in generative AI and probabilistic graphical models. I have extensively worked on models such as Latent Dirichlet Allocation, Graphical Models, and Markov Models. Furthermore, I have contributed to the development of a generative topic model that can disambiguate words, and designed a record linkage system that adapts to new data with high precision.

Future directions for my research include improving the interpretability of machine learning models, a crucial step towards their wider adoption in various industries. Additionally, I aim to further explore probabilistic methods to create more robust models capable of handling uncertainties inherent in real-world data. I believe that this research direction will allow me to produce cutting-edge results, and will also align with the growing demand for transparent, reliable, and efficient AI models in numerous industries.

In summary, my research endeavors are geared towards pushing the boundaries of machine learning and AI, with a keen eye on their practical applications in the real world. As an AI scientist and engineer, I am motivated by the challenge of making AI technologies more accurate, efficient, and useful to society. I am excited to continue my work in this direction and look forward to the many challenges and opportunities that lie ahead.