Suhas Manangi

Product School, Try Exponent, and Product Allinace are good resources for PM interviews prep. Later is a good question. Interesting idea. I don't know of any, but it so interesting that someone should be offering it. Perhaps they might have rolled into certification or cohort courses with live projects!

Being a good PM helps becoming a good manager of PMs, but is not a sufficient condition. I have seen below 3 as top challenges/opportunities unique to GPMs: 1. Deligating, and trusting your direct report PMs to care about Customers as much as you do, if not more. 2. Providing saftey net for PMs to fail fast, learn, and iterate, but as well the essential framework on lowering the cost of failure to ensure contribution to business impact. 3. Knowing that PM skills are not hard to aquire, but takes time. Coaching the team on specific PM skills need persistence and patience. It is not like you launch a product, and you see a metric go up instantaneously.

First PM in a company! I have not done it, nor have anyone in close network to have a good understanding. My guess is that they have to establish right roles/responsibilities on what to carve out from the CEO. Perhaps focused on scaling up product for next million users (or take on next set of enteprise clients), or execution focsed. Do take this with a grain of salt as I am guessing based on when should a CEO hire their first PM.

Top 3 traits that makes a Good PM a Good AI PM: 1. Understand foundational ML tech concepts and having used them to make product decisions. For eg: Statistical Regression, Causation vs Corelation, AUC, P/R, Features vs Labels, Feature distribution, Model Training, Model drift and auto training, etc 2. Aware of potential bias and fiarness need in ML solutions they have launched in the past. Having used model observability and interpretability to explain the model output for their product corner cases. 3. Ability to scale up product decisions going from single global configuration, to customized per user segment, to fully scalled 'personalized product expereince for each and every user'.

5 years from now, likely there is going to be no difference between a Traditional PM and AI PM. AI is going to be used/present in all products. I see "Traditional PM role" as a foundational one to have, upon which one can grow to become a good AI PM. Good Traditional PM with aptitude for tech and data science is likely to do well as a good AI PM. Taking a Udemy course on basics of AI/ML, and applying to every day PM job will be a great start.

Trust & Risk is a specific domain, so generalist PM would need to pivot to become a domain PM. This will require launching solutions across multiple years. Fraud evolves every year, but over 90% still remains the same decade long type. Payment Risk is a good stepping stone. Balacing security with usability (frictions to stop bad actors) which translates to doing a trade off between fraud loss and revenue loss is a new skill PMs have to develop. 

In addition to core business metrics that are improtant for a product success, below are the additional ones AI PMs obsess over to ensure the success upon launch doesn't regress over time. 1. Precision and Recall 2. False Positive 3. Model quality monitoring metrics based on where is the risk to business (feature quality, score shift, re-training frequency, etc)

PMs are generally categorised into B2C (Consumer), B2B (Enterprise), Platform, and Product. PMs role is generally 2 of these 4 things. Within this one can also be generalist PM vs domain PM vs growth PM vs scale PM etc. AI PM is the same. One additional categorey can be about building ML Ops platform but I am not convinced one needs a PM for it, or can't be fit into one of the 4 categories described above. PMs working on Alexa can be AI PMs, but not necessarily TPMs.