From data science to product management: the journey of a woman in AI
It is no secret that the majority of people working on AI and machine learning projects today are men. Gender balance in AI conference lineups, in Silicon Valley and around the globe, has been noticeably poor. One study found that only 12 percent of machine learning researchers are women, a shocking statistic if we consider that AI has the power to reshape our society.
In my own journey as an immigrant Latina working in tech, I have heard many times that “You can’t be what you can’t see,” and this is true for all of us. We rely on examples, stories, and images that help us believe in our potential and the people we can become. If we want to close the gap and have more women working in deep tech fields, we need to make sure that the current generation of young women grow up with leading women as examples to be inspired by.
We recently partnered with WomenOfAI.org, a global nonprofit organization promoting women leaders in AI, to support their mission to inspire more women to build their careers in AI and machine learning. As part of this collaboration, we hosted an event on April 22 on “AI for the Digital Food Experience,” featuring Lotem Peled-Cohen, a machine learning product manager at Whisk.
Asking the hard questions
Lotem Peled-Cohen is a machine learning product manager at Whisk, a food AI startup recently acquired by Samsung NEXT. In her role, Lotem leverages her domain expertise and understanding of machine learning to solve the problems with the largest business impact.
Lotem’s passion is in shining an AI light onto the world of product management. She discovered her passion for machine learning while getting her bachelor’s degree in information systems in Israel. After that, she received a master’s degree in natural language processing (NLP) and became a data scientist. Even in her early days as a data scientist, Lotem’s greatest excitement was always based around answering hard, business-related questions.
AI is a tech-heavy domain that also invokes a lot of sentiment in users; therefore it calls for very specific product questions.
“I felt that there was a gap in the product management of machine learning, no one was asking questions like: ‘How do we monetize this product, which is built on ML models? Based on the intelligence level of the solution? Or, for example, what do the users feel about incorporating such an AI model? How do we market such an ML based product?’ and these kinds of domain-specific questions. I was very excited about the challenge of communicating ML to others and answering these questions, so I went on to become a product manager that specializes in [machine learning]. I then found myself at Samsung NEXT and I’m very happy that I’m here.”
You might be asking yourself what is the main difference between a “product manager” and a “machine learning product manager”? In Lotem’s words:
“Traditional product managers are essentially enabling developers, whether they’re designers or engineers, to create specific features for a bigger product roadmap. With machine learning projects, because it’s different from traditional software and very intense on data and modeling, the actual process and people involved will include not only engineers and designers but also a data science team. First you need to really invest in data and then you can build a true AI and truly successful ML models on top of it.”
If you are considering a career change and want to learn more about the main differences when you shift from classic product management to product management based on machine learning, you should read this article.
Shifting your career
Lotem has been able to move from data science to product management in less than a decade, but starting your career in product management or data science is not as easy as it may look. In her opinion, getting your first job as a product manager might be as difficult as getting your first job as a data scientist.
“I just want to say to anyone who wants to be a data scientist and they’re currently looking for a junior position, don’t despair! You will get to that position at a certain point. I just want to give a little bit of hope to people who want to work as a data scientist and don’t have a master’s or a PhD,” she said.
Currently, the market is in a place in which even for a junior data science position, companies are requiring very high-level academic degrees. She believes that this will change in the following years because there will start to be a stronger line in the sand between research and product. The same applies to product management as well. If you are young, fresh out of college, and interested in getting your first job as a data scientist, a great tip is to focus your time on building a portfolio.
“Go online, go to multiple hackathons and competitions, and have projects behind you to prove your knowledge. Just show them that you’re a good engineer. Don’t forget that to be a junior data scientist, you have to be an excellent developer.”
She also noted that the type of machine learning required for engineering and productization is completely different from the machine learning required for innovation research or writing academic papers. If you are a data scientist or a traditional product manager switching to a machine learning product management role, keep in mind that ML requires a more experimental approach since it involves learning from data and making decisions based on data instead of following a set of human rules like in traditional PM and software engineering. A great ML product manager is normally an expert in data, using theories and models on top of UI design to solve customers' problems.
Lotem has been a machine learning PM at Whisk for almost a year now, and she is finding this position unique and challenging.
“I get to dive deep into user data, spec out ML products that would make our users happy, work with designers and engineers on bringing these products to life, and even help with marketing initiatives. I love leaving the ML tech to our talented data team, while I focus on communicating ML decisions and dilemmas to others. Actually, what attracted me to join the company in the first place was the fact that Whisk is actually all about the data. Our team members know that first you need to really invest in data and then you can build true AI on it, and truly successful machine learning models on top of it.”
Lotem is happy to mentor any fellow member of WomenOfAI.org thinking about becoming a data scientist or ML product manager. Feel free to reach out to her via her LinkedIn page!
Since 2013, Whisk has been powering half a billion connected food experiences per month across millions of online recipes with many of the world’s leading food publishers, brands, and retailers. Whisk’s Food Genome is an intelligent deep learning-based natural language processing algorithm that maps the world’s food ingredients, their relationships, their properties (nutrition, perishability, flavor, categories) as well as offering food purchase options. It’s a combination of this deep ontological understanding of food with massive data about user behavior that drives accurate, smart results from any source, at scale. To learn more about Whisk, go to https://whisk.com/.
