OTC CatchUp #176 Summary

100ms is a platform that provides video conferencing APIs for developers to build video conferencing applications. They have many features like recording, streaming, and a lot of other things that are needed for a video conferencing platform.

Harsh Kapadia thought that since it is a video infrastructure, the amount of work would have been deterministic, but Aniket Kadam mentioned the fact that their work isn’t still complete since they keep adding new features and functionalities based on customer requirements.

Aniket Kadam talked about how an average user mostly reads compiler errors instead of following the documentation. So they push most of the information in the compiler errors to make the users aware.

Aniket Kadam suggested that Android seems like a bunch of completed Jira tickets, and less like a polished consumer product, whereas iOS seems to be a bit more coherent and polished.

Most people suggested that he should choose what he likes based on exploration and not what the curriculum is pushing him toward.

He described the working of the tool including the usage of presidio and how it generates unique fake data for the columns in the Excel file.

He also talked about Busybox, which is a software suite that provides several Unix utilities in a single executable file.

We also discussed victim cache and cache structure.

Dram structure, its density and how the memory controller refreshes the contents of dram in an interval so that the banks don’t lose their charge.

He also talked about the structure of sense amplifiers and how they are used to read the data from the dram cell.

Harsh Kapadia shared an interesting article about CPU caches.

Pranav Dani shared Cache Associativity Handout to understand the concept of cache associativity easily.

Jaden furtado shared a book for Computer Organization and Architecture by William Stallings.

He also then talked about TableNet, which is a deep learning model for end-to-end table detection and tabular data extraction.