Chandrakanth R Kancharla

My interest in applied research is what prompted me to pursue my current PhD on 'Machine learning for resource constrained edge'. I design and investigate domain invariant, self-supervised and unsupervised models, owing to the scarcity of labelled and heterogeneous data in real life. Python, C++, Git and other python packages related to machine learning and data analytics are my tools in this quest.

Before the PhD at KU Leuven, I was a co-founder of an R&D startup in India. I've learnt a lot about managing products and a team. Designed and developed various proof of concepts for the clients during my tenure. Proficiently used Computer Aided Design (CAD) tools, Python, C++ and Android studio, and, web development tools in limited capacity. Notable results of our collaborative effort are:
- We were named one of the "Smart 50 startups transforming India" by the Department of Science and 
Technology, Government of India, and IIM Calcutta.
- We were also in the top 20 global startups for the Samsung global acceleration program 2018, Korea.

A keen interest in technology and a curious mind propel me forward. Always up for a coffee, feel free to message over LinkedIn.

Values: Invent or innovate, make sure user's need is met in whole.

Data fascinates me, and the practical challenges in collection and analytics pique my interest even more.

Data scarcity is a problem in many industries and use-cases in manufacturing. First, conditional heterogeneity is absent in data, and second, the collected data is mostly unlabeled. The bandwidth limitations of Industrial IoT networks, as well as the need for low latency fault diagnosis, add to the complexity.

Given these challenges, my research focuses on domain adaptive/generalizable Machine Learning models that can be implemented on resource constrained edge devices (typically a micro-controller). While I discovered the use of Auto-encoder latent features as generalizable features for rotating machine fault diagnosis, I am currently expanding on this concept by self-supervision to adapt to complex unknown conditions.

• Ongoing PhD in Computer science, specifically focused on domain invariant machine learning for resource constrained edge devices.
• Master's in Advanced control and robotics