I am currently designing a university course on ML-Product Management.
The module's focus is to teach my students that when it comes to building efficient and effective solutions the problem-analysis comes first and the ML-side is just a part of the overall challenge. To that end the student teams collaborate with industry and government institutions to develop solution designs and prototypes.
Some of the best practices we apply are based on my experience as a CSO in the contract intelligence space:
- solution scoping based on task complexity, open-ness of world model, and dynamics of environment
- business metrics first, ideally correlate ML-metrics to business metrics
- fast iteration and evaluation on the model level
- fast wire framing and prototyping to expose the ML-based functionality to end-users
I am curious how other's approach this task? What is your approach, what literature do you recommend?
0 comments
—
Ask HN: Best Practices ML-Product Management · HackerTrans
The module's focus is to teach my students that when it comes to building efficient and effective solutions the problem-analysis comes first and the ML-side is just a part of the overall challenge. To that end the student teams collaborate with industry and government institutions to develop solution designs and prototypes.
Some of the best practices we apply are based on my experience as a CSO in the contract intelligence space:
- solution scoping based on task complexity, open-ness of world model, and dynamics of environment - business metrics first, ideally correlate ML-metrics to business metrics - fast iteration and evaluation on the model level - fast wire framing and prototyping to expose the ML-based functionality to end-users
I am curious how other's approach this task? What is your approach, what literature do you recommend?