mcgrathpm11

what problem did the model solve? eg, did it predict something, optimize?

I've found that communicating results of an ML model to non technical stakeholders is as simple as:

1) we are building the model to predict X 2) the model currently predicts X with Y% accuracy 3) I am doing X/Y/Z, eg feature engineering, to improve accuracy

users, showing you learned from user feedback to iterate on the project

I worked at Netflix for several years, where they actively practice paying ‘top of market.’

This is an oft misunderstood tactic, but in action means: 1) Yearly, employees are asked to seek evidence for their top of market compensation. 2) employees do two things as a follow up - They either interview elsewhere to get competing offers to establish ‘top of market,’ or they feel fairly paid and do nothing.

Having worked at a few large tech companies that practice the typical equity grant and vesting structure designed to maximize retention, Ive never seen a more effective tactic to completely remove the concern of ‘am I paid enough’ from the employee. As a result:

1) The really high impact people are paid extremely well, as Netflix always overbids. 2) The average performers either go elsewhere or retain without jockeying for greater comp and ownership.

The proactive seeking top of market is a good feedback loop to pay for/reward impact and to build trust with your team, and is something I still practice with my teams outside of Netflix.

This was back in 2015, so some things may have changed. Regardless, the net to the business has compounding returns. At an average company:

1) Most employees do not pay for their salary in business impact. 2) Top performers drive outsized impact relative to their compensation.

If you find a way to mitigate 1), you can pass the savings into 2), protecting and retaining your highest value employees.

While other factors do obviously matter, and comp isn’t everything - removing comp from the hierarchy of needs entirely dramatically changes the mindset towards ownership of work and outcome.

Congrats, Paras. I've been a huge fan of VWO and fairly recently built somewhat of a competitor - evolv.ai. I've referenced your sales assets, training videos, and I have sent people your direction when they weren't a fit for our offering.

The primary point of the article is that dog photos lead to more 'introductions' between two users, while cat photos lead to more threaded messages - which likely lead to more in person meet ups.

These 'fantasy' requests are equal parts hilarious and ridiculous. People are weird.