Technical Learning: You have not thought it through!
After crossing the midpoint, the pace of the internship accelerated. To receive buy-in for our approach to build a speaker identification system for Samsung TVs, the team went full steam ahead on building a proof of concept (POC). The key challenge with building a POC was the hardware constraints… well because it would need to work on a TV!
“Even the most advanced televisions today, don’t have the level of compute power offered by a GPU or multi-processors.” — My Manager
Hence, to build an on-device solution, we’d need to strip the model to its bare minimum and use algorithm acceleration techniques to build a feasible solution. This experience taught me the importance of considering all eventual constraints, even when working in the open-ended domain on research.
The Best Learning: Community makes it all worth it
Given how close I was to the other fellows in Asia, it made perfect sense to visit them to finish off the internship. I planned two visits:
I had spent the first half of 2019 in a foreign place — first on exchange, and then for an internship. Being uncomfortable forces you to adapt, and learn a lot about yourself, but often this process is solitary. Meeting the fellows, and feeling the strong sense of community that was established in early May, made the (fairly tough) experience, worth it. I am grateful for the other fellows that opened up, and shared their trials and tribulations with me. I’m also thankful they allowed me to do the same.
If you enjoyed reading this post, and are interested in joining the incredible Cansbridge community, please apply here in the fall! William Yu, the founder of Cansbridge also visits Canadian campuses. You can also contact me at shekharkumar.com if you have any questions.