Shobhit Varshney

There is a continuous spectrum from (1) Creating AI to (2) Applying AI. The job market for Applied AI is massive, where you need to have a major in a particular domain, and minor in AI. Last year, I mentored a young mom who had been working at Starbucks for 14 yrs, and now that her daughter was in school, she was taking a 1yr program in AI to shift her career. For people like her who are transitioning or new to AI, they should lead with their domain expertise, and get sharp on how AI could be applied to their industry/role/domain. She recently started at Target in their Analytics division for retail experience. She is making operational changes to capture new data points, and optimizing how AI is making decisions. She is crushing it.

So far the discussion here has been geared towards the AI SMEs who are optimizing the costs of choosing/deploying a particular LLM. Looking at the overall enterprise landscape, you need to abstract the complexity of choosing/ optimizing/ governance of LLMs away from the business teams. We see huge impact when we deliver an AI platform where app teams (across marketing, sales, IT, etc.) focus on business logic, and the central AI team creates centralized, robust, scalable and secure assets. There are huge inefficiencies in how various teams pick their models. That needs to be a specialized function that considers security, cost, latency, IP etc. when choosing and deploying models.

This is something you need to lead by example. Every time there is a major event/ conference/ announcement, we have a watch party or have the team frantically pinging on slack/text/etc. As we cover the event, we usually start collaborating on a debrief box note, and have a junior team member present it at the next team meeting. Or leverage that for a LinkedIn post with a personal take on the announcements.

If you can't respect your personal time, you can't expect others at work to do the same for your time. My work calendar is permanently blocked for all year for late evenings and super early mornings. My admin needs to clear it by me if I get pulled into any meeting before 7:30am or after 5:30pm when my kids come home. I use that as a forcing function to build a backup for after hours, so eventually that team member could take over my job while I am out on vacation or want to free up to move to a different role.