Using Synthetic Data to Train AI Models | Ido Halevi | Director of Product Management at Data Gen image
The Building Blocks Podcast
Using Synthetic Data to Train AI Models | Ido Halevi | Director of Product Management at Data Gen
Using Synthetic Data to Train AI Models | Ido Halevi | Director of Product Management at Data Gen
Ido is currently the Director of Product at Datagen, a company focused on providing high-performance synthetic data, with a focus on data for human-centric computer vision applications In 2022, Datagen closed a 50 million dollar Series B funding round and has been working with clients across the world like Nvidia, Google, IBM and Intel. We dive deep into Ido's vast experience as a Product Manager and the nuances of using Synthetic Data to train AI models. We go through how Datagen goes about sourcing its data sets, how they go about curating their initial training set, how the evaluation of the AI models and the future of the platform. Learn more about Datagen: ⁠https://datagen.tech Guest: Ido Halevi LinkedIn: https://www.linkedin.com/in/ido-halevi-381603147/ Connect with Daivik: ⁠https://bento.me/daivik⁠ Follow the Podcast: ⁠https://bento.me/thebbpod⁠ Topics 0:00 - Intro 1:30 - What brought him into Product Management? 7:45 - Talking about Datagen 14:50 - Talking about the benefits of Synthetic Data 22:40 - How does Datagen source data for its own models? 26:05 - How does Datagen decide what features to let its customer control? 31:30 - How do they evaluate Synthetic Data Models? 37:09 - AI Startups 43:40 - Building the platform 47:40 - Wrapup ----- In the upcoming episode of our podcast, we're thrilled to host Ido, an industry leader in Product Management and currently the Director of Product at Datagen, a pioneering startup focused on providing high-performance synthetic data, with a special emphasis on data for human-centric computer vision applications. Ido's journey began at Intel as an Engineer before venturing into Product Management at Logz.io and then serving as a Senior PM at Explorium before finally joining Datagen. Outside of work, he actively mentors at Scale Up, an organization addressing human capital challenges in the hi-tech ecosystem, and also contributes to Max initiative, an educational impact entrepreneurship program. During the podcast, we'll delve into Ido's story, exploring what led him towards Product Management and how he has navigated his career decisions. We'll also discuss his invaluable experience building products at Datagen, a platform enabling users to generate synthetic data tailored for their machine learning models. We'll learn how Datagen addresses the critical role of data in model training, particularly in the challenging domain of Computer Vision. Ido will shed light on how they source data for their own models and the design considerations for their Neural Nets in generating realistic synthetic data. We'll explore the intricacies of evaluating GAN-based models and the compelling advantages of synthetic data over real-world data, from cost-effectiveness to mitigating bias. The future of Datagen's platform will be a focal point, as we'll uncover how they strike a balance between offering user control while maintaining a generalized self-serve platform. Ido will share some recommended reading and offer insights into the next steps for himself and Datagen. Don't miss this enlightening episode that unravels the wonders of synthetic data and its transformative potential in the world of AI! 🎧 #ProductManagement #AIInnovation #Datagen #BuildingBlocksPodcast #AI #ML #data #syntheticdata #computervision #product #technology #startup #podcast #founders #productbuilding
00:00:00
00:00:01
0 Plays
1 year ago
Ido is currently the Director of Product at Datagen, a company focused on providing high-performance synthetic data, with a focus on data for human-centric computer vision applications In 2022, Datagen closed a 50 million dollar Series B funding round and has been working with clients across the world like Nvidia, Google, IBM and Intel. We dive deep into Ido's vast experience as a Product Manager and the nuances of using Synthetic Data to train AI models. We go through how Datagen goes about sourcing its data sets, how they go about curating their initial training set, how the evaluation of the AI models and the future of the platform. Learn more about Datagen: ⁠https://datagen.tech Guest: Ido Halevi LinkedIn: https://www.linkedin.com/in/ido-halevi-381603147/ Connect with Daivik: ⁠https://bento.me/daivik⁠ Follow the Podcast: ⁠https://bento.me/thebbpod⁠ Topics 0:00 - Intro 1:30 - What brought him into Product Management? 7:45 - Talking about Datagen 14:50 - Talking about the benefits of Synthetic Data 22:40 - How does Datagen source data for its own models? 26:05 - How does Datagen decide what features to let its customer control? 31:30 - How do they evaluate Synthetic Data Models? 37:09 - AI Startups 43:40 - Building the platform 47:40 - Wrapup ----- In the upcoming episode of our podcast, we're thrilled to host Ido, an industry leader in Product Management and currently the Director of Product at Datagen, a pioneering startup focused on providing high-performance synthetic data, with a special emphasis on data for human-centric computer vision applications. Ido's journey began at Intel as an Engineer before venturing into Product Management at Logz.io and then serving as a Senior PM at Explorium before finally joining Datagen. Outside of work, he actively mentors at Scale Up, an organization addressing human capital challenges in the hi-tech ecosystem, and also contributes to Max initiative, an educational impact entrepreneurship program. During the podcast, we'll delve into Ido's story, exploring what led him towards Product Management and how he has navigated his career decisions. We'll also discuss his invaluable experience building products at Datagen, a platform enabling users to generate synthetic data tailored for their machine learning models. We'll learn how Datagen addresses the critical role of data in model training, particularly in the challenging domain of Computer Vision. Ido will shed light on how they source data for their own models and the design considerations for their Neural Nets in generating realistic synthetic data. We'll explore the intricacies of evaluating GAN-based models and the compelling advantages of synthetic data over real-world data, from cost-effectiveness to mitigating bias. The future of Datagen's platform will be a focal point, as we'll uncover how they strike a balance between offering user control while maintaining a generalized self-serve platform. Ido will share some recommended reading and offer insights into the next steps for himself and Datagen. Don't miss this enlightening episode that unravels the wonders of synthetic data and its transformative potential in the world of AI! 🎧 #ProductManagement #AIInnovation #Datagen #BuildingBlocksPodcast #AI #ML #data #syntheticdata #computervision #product #technology #startup #podcast #founders #productbuilding
Recommended