Suggestions
Anastasia Martynovitch
Product Manager at Zoox
Anastasia Martynovitch is a seasoned Product Manager with a six-year background in enterprise SaaS and B2C companies. She notably served as the founding PM at Surfline's e-commerce platform, enhancing user engagement by connecting millions of users with surf products from partner brands.
Her expertise lies in navigating the dynamic environments of startups, where she excels in collaborating with cross-functional teams to drive product strategy and roadmap. Anastasia is known for her hands-on approach to data analysis and her willingness to engage in the technical aspects of Product Management, including coding.
Anastasia's educational background includes a Bachelor of Business Administration focused on Business Administration and Management from Wilfrid Laurier University. She also pursued studies in Computer Science through The Web Developer Bootcamp from Udemy Academy and a minor in Computer Science from the Lazaridis School of Business & Economics at Wilfrid Laurier University.
Throughout her career, Anastasia has held various roles such as Product Manager at Zoox, Senior Product Manager at Turvo Inc., and Associate Product Manager at Prenetics. She also ventured into business development, sales, and co-founding a startup. Her diverse experiences have honed her skills in product development and management.
In her leisure time, Anastasia enjoys cycling, pottery, and exploring the Bay Area. She is open to restaurant suggestions and engaging in conversations about interests like Stardew Valley.
Highlights
I’m happy to share that I’m starting a new position as Product Manager at Zoox! Zoox vehicles don't have a steering wheel because they're purpose built to be self-driving. So excited to be part of the team getting Zoox ready for a public launch!
Why might accuracy not be an good way to assess an ML model?
Well, what if a model is trying to identify heart disease among 1,000 people, with only 10 actual cases of the disease?
If the model predicts "no heart disease" for every person, it would still achieve a 99% accuracy by correctly identifying 990 people without the disease, but it won't do it's core job of detecting any actual cases.
A better metric is called "recall" which is the % of actual heart disease cases that the model correctly identifies.
This is a condensed version of what I'm learning in this ML course! I love discovering logical fallacies like this and learning how to avoid them.
Here's the link if you're interested in the course: https://lnkd.in/eg9cZJsM
Let me know if you end up checking it out—I'd love to have a buddy to work on this together.
