Throughout my career of six to seven years, I have specialized on artificial intelligence and machine learning product management. I joined an established company as an AI/ML PM after breaking into the field at a big professional services firm and working at an F50.
Let’s speak about how many companies, my own included, use the phrases artificial intelligence (AI) and machine learning (ML) as if they were magic bullets when interacting with clients. In order to become knowledgeable enough to reject as ridiculous the request to “integrate AI into our product,” I need to enroll in several programs, read a lot of literature, and attend innumerable interviews.
I agree that it’s easy to abuse AI, but I also believe that it’s your responsibility as a product manager to inform stakeholders and end users of the limitations of your product. Having clear expectations is particularly important when dealing with users.
Which technical tools do you have and use most frequently as an AI PM that you believe are absolutely necessary?
Just like using tools like Balsamic, figma…etc to sketch up low fidelity mock ups for designers. I think working with data goes a long way as a AI PM working with data scientists.
Being able to tell a story with sample set of data helps a lot. Python is great overall. I also picked up tableau to visualize the data I work with.
Heck, even MS Excel has worked for me in the past to do data analysis and visualization (if someone isn’t interested in learning Python…etc).
I don’t know if any of this is “absolutely” necessary to be a AI PM though - just found it super helpful for me personally. I think the basics of being a PM still stands as an AI PM. Focus on the customer, problem and value to the business i.e The what, why and when - let your technical team worry about the how.
As an AI PM, I believe it is absolutely necessary to have a strong understanding of programming languages such as Python and R, as they are commonly used in AI development. Additionally, knowledge of machine learning frameworks like TensorFlow and PyTorch is crucial for implementing AI algorithms effectively. These technical tools enable me to effectively communicate with developers and make informed decisions regarding the integration of AI into our product. Furthermore, familiarizing myself with natural language processing (NLP) libraries, such as NLTK and SpaCy, has been extremely beneficial in understanding and working with textual data. Having a grasp on cloud computing platforms like AWS or Google Cloud is also essential in efficiently managing AI infrastructure and deploying models at scale. In summary, possessing proficiency in these technical tools not only facilitates collaboration with developers but also empowers me to steer our AI projects in the right direction.
That is very helpful information. I appreciate you answering my question and I hope you achieve great success in your career.
Thanks to you too for taking the time to reply me specifically. Very thoughtful of you.
What industry are you in?
What were the types of products? recommender systems? Classification systems, etc.?
How big was your tech team + composition?
Did you do 0-1, or did you build on an established product?
Did you work on the infrastructure layer, the HCI components in the front end? Both?
Financial Services → (Felt like I liked the domain of FinTech and wanted to stick to it)
Classification problems/products mainly, but also worked on forecasting problems/products
Generally the size/composition has been very similar with the exception that my second role was a primarily offshore team. I typically have worked with 3-5 Data Scientists, a designer, 2-3 Devs for any UI work and 1 Data Engineer. The team also then interfaced with a larger “platform” team or engineering team that maintained the wider architecture/production environments.
0.5-1, 0-1. My roles have been around bringing new products to users. So primarily 0-1 experience.
Thanks for that- very informative.
I’m in a similar position in FinTech (more similar to a PM for a hedge fund) and I am working across platform + UI together.
What do you like most about your job and what do you dislike most about your job to date?
Your description about AI/ML PM piqued my interest. How does your product design process vary from that of PMs who do not use AI or ML? At this point in time, I believe that the majority of goods will include ML in some way. Is the ability to direct engineers to employ a certain algorithm the reason you refer to yourself as an AI/ML PM? On the other hand, is it mostly about getting the current AI/ML model to work better for you?
While the ideas behind design remain constant, the actual procedure varies. Not every issue can or should be handled with AI, hence AI project managers need to know which challenges are AI problems. Data and the processes an engineer will use to collect, clean, and work with it are also common topics for project managers to be familiar with.
When you’re planning and constructing a model, keep in mind the steps that make up its development lifetime. In the end, it affects product positioning, resource management, trade-off determination, and roadmap building. In comparison to more conventional software solutions, I would argue that AI/ML offerings carry a much higher degree of uncertainty.