In this episode of Unboxing AI, I host Idan Bassuk, the VP A.I. at Aidoc, to chat about computer vision, NLP, and AI explainability in the medical field. During the episode, Idan discusses the challenges that should interest anyone who’s building an AI team for scale. For example: what type of roles does he have on his team? Is it engineering first or data first? Does the real world of production resemble the rich academic research in the medical space?

TOPICS & TIMESTAMPS

1:52 - Going from terror tunnel detection to tumor detection

4:00 - Aidoc’s medical imaging AI product in a nutshell

9:09 - Main challenges for Aidoc

10:50 - Data variability in the medical field

14:18 - Explainability in medical AI

17:02 - Incorporating SoTA

19:15 - The state of academia

21:30 - Data-centric AI

23:49 - AI org structure at Aidoc

27:08 - Building test sets

29:19 - Using NLP to accelerate algorithm development

31:18 - Finding the right data

35:53 - How choosing the right annotation method affects accuracy

40:43 - Bringing it to production - team coordination

45:11 - The importance of clean code and code review

51:43 - Hiring with extreme transparency

54:46 - On the role of AI software engineers

57:08 - CI/CD and reproducibility

58:32 - Working agile in AI

1:01:38 - Planning with uncertainty

LINKS AND RESOURCES

Idan Bassuk LI: linkedin.com/in/idanbassuk

Aidoc: aidoc.com

Stanford cs231n course, mentioned @56:46:

Course Lectures: bit.ly/cs231n-Karpathy

Course Homepage: cs231n.stanford.edu

GUEST BIO

Idan was Aidoc's first employee. He started as an AI Algorithm and Software Engineer. Today he leads the A.I. Group, a 90 people group that concentrates all the efforts required for A.I. from dataset development, through algorithm research and engineering, and up to deployment and continuous monitoring in production at a scale of over 500 medical centers worldwide. Before joining Aidoc, Idan served for 10 years in the Israeli Defense Force. He started in the elite technological Talpiot course and later served as a team leader in a special operations unit. Idan finished his service as a Head of the Technological Section, leading defensive technological projects which were awarded Israel’s most prestigious defense award, for the success in detection of tunnels crossing the border to Israel.

ABOUT THE HOST

I’m Gil Elbaz, Co-founder and CTO of Datagen. In this podcast, I speak with interesting computer vision thinkers and practitioners. I ask the big questions that touch on the issues and challenges that ML and CV engineers deal with every day. On the way, I hope you uncover a new subject or gain a different perspective, as well as enjoying engaging conversation. It’s about much more than the technical processes – it’s about people, journeys, and ideas. Turn up the volume, insights inside.

FULL TRANSCRIPT AND MORE AT UNBOXINGAI.SHOW:

https://unboxingai.show/podcast-item/saving-lives-with-deep-learning-robust-infrustructure-idan-bassuk-aidoc/