Krisp reposted this
Zoom is one of the largest Voice/Video platforms in the world. Ever wondered how Zoom builds Voice AI internally? In this interview, Tommy Gaessler, Lead Developer Advocate/Engineer at Zoom and I do a deep dive into how Zoom builds Voice AI. Here’s what stood out to me most 👇 1) Zoom has an API and SDK-first mindset 2) The SDK is a core technology 3) Zoom’s SDKs see over 55,000 weekly installs across various package managers and platforms 4) Zoom has a world-class AI team led by CTO Xuedong (XD) Huang, previously CTO at Microsoft Azure AI. 5) Zoom's platform approach aims to open as many building blocks as possible for various use cases. 6) Zoom takes a federated approach, allowing developers to integrate AI models within Zoom’s infrastructure, providing maximum flexibility 7) Prioritizing features in product management is a challenge due to the platform's scale. 8) Zoom's approach to problem-solving is taking a few steps back to fully understand it before solutioning. 9) Typically Zoom prefers building its own technology rather than buying or acquiring products. 10) Many open-source technologies are not great, prompting the need for better proprietary solutions that will benefit users. 11) By owning and continuously improving core technologies, Voice AI companies are more likely to become category leaders and own the use case. 12) Zoom is exploring AI use cases for enhancing sounds, like hearing a heartbeat better in telehealth. 13) A new voice AI inside Zoom Rooms allows identifying the primary speaker and focusing the video on them 14) Zoom was built on customer feedback and it will continue to drive their roadmap and innovations in the coming years. Tommy, thanks for your time and insights 🙏 Full interview here 👉 https://lnkd.in/egii8yJc