My lab [Innovatory for Cells and Neural Machines, ICAN] is innovating at the forefront of applied machine learning and data engineering. Given that my first (independent) grant ever was from NIH [NIH R01, NIAID] at the interface of machine learning [for the microbiome] and data engineering [to harden the backend of the largest microbiome data analysis pipeline, MG-RAST], it is perhaps no coincidence that my lab resides at the nexus of the two seemingly disparate areas comprising of machine learning and data engineering.

At ICAN, machine learning is primarily applied to genomics problems. One specific focus has been on ML for epigenomics, which is essentially the dynamic layer of chemical modifications atop the genome. The epigenome is magical because you can change it by doing the right things---lifestyle decisions, the food you savor, and who knows, perhaps even the thoughts that emanate from your being! It is dynamic and can interactively affect the computation [unified chord or Aikyatan in Sanskrit] of the cells that host the code of life.

In contrast, data engineering primarily manifests itself in my urge to want to engineer the algorithms to be resilient in the wild---whether it be the heterogeneity and Byzantine attacks in federated learning or the democratized IoT sensing in the large livestock ranches or ecologically-aware digital farms. You want to onload intelligence onto the miniature sensors so that the machine learning pipeline can be instantiated right at the sensors, followed by intelligence at the edge, and finally at the cloud servers, hosted at the data centers. At these data centers again, you may want to disaggregate compute, storage, and memory to improve data center efficiency and be able to rightsize computation for the workloads presented by the clients.

With these nuances of my journey at the exciting boundaries of biological engineering, IoT, and computer engineering, ICAN is now podcasting off and on @Non-linear Computation! Podcasting is my way of directly talking to students and was initialized at the foundational and advanced (Capstone-style) applied machine learning courses I teach at Purdue. I hope to nurture it some because I love my tech journey and I look forward to sharing my excitement with you.

Spotify Apple

Podcast Notes

Explanatory Notes for the Podcaster