#datascience

I recently started seeing few posts on LinkedIn about some people posted about being laid off.. like L3 L4 L5s in engineering, data science etc. anyone noticing anything? I think mgnt figured out a way of staying out of news this year breaking habit of yearly mass layoffs.

When the dusts of current RIF and ReOrg settle, architects and the data scientists are the two most costly and counter-productive roles that needs the closest scrutinies.

The existence of those helicopter architects is the single inhibiting cohorts to an engineering driven culture. If the premise of those tenured architects is to guide the weak engineering teams, it’s not working and will never work (that’ why you don’t see the architect role in big tech such as Google and Meta):

• If you keep the architects away from the engineering team like it is today, their lack of current and hands-on knowledge, and their lack of affinity to the day-to-day work on the one hand, stretches a tension with their assumed authority on the other hand. We find ourselves wasting cycles and energy convincing and compromising with them on a good day and misled/delayed on the bad days. We consider those architects good if they stay away most of the time, reverse engineering by themselves or asking us to produce a few pretty diagrams from our finished products periodically, and don’t try to put their dirty fingerprints on everything we do.

If you pull those architects Gods down from the FAE heaven and embed them into the engineering teams, their unwillingness to do the dirty chores and their proud refusal to assimilate will create tension between themselves and rest of the team like oil and water. Neither party will be happy.

That leaves the only option which is to reduce the architect role dramatically if not demolish it altogether: (1) Keep only a few true architects in FAE who either (a) looks cross-functional for duplication and consolidation opportunities or (b) possesses niche knowledge and skills such as security, internationalization or accessibility. (2) Fire the rest or demote them to principal (or just give them the VP title) level IC and disperse them into individual engineering teams.

With the deadweight architects out of hand, we may start growing the engineering team and culture by trusting engineers with the architectural decisions in a collective fashion among the junior, senior, principle and tech leads of a team.

Now let’s turn our attention to data scientists. WARNING: they’re so much worse than the architects!

The complete AI ignorance of the upper managements makes themselves easy targets to the scammed by Fidelity’s fake data-scientists (compared with those who can build GPT):

• They’re paid at least one level higher than engineers yet what are they doing these days? Developing chatbots by calling vendor APIs or downloading models from HuggingFace. What entails in developing chatbots or the fancier agents? (1) Calling APIs, (2) developing the chat GUI and (3) Crafting LLM prompts. Well, software engineers are better calling APIs and developing GUI, and non-technical business domain experts are better at crafting LLM prompts. Both do a better job significantly cheaper. AI has been demoratized to a point where a high-school drop out may do a better job than an Ph.D. who don’t continuously learn.

Why don’t they train foundational LLMs that utilizes Fidelity’s private data, like Bloomberg, Captital One or Morgan Stanley? They can’t. The whole data scientists community from top to bottom are outdated. They’re stuck in the old traditional machine learning paradigms of regression, decision trees and scikitlearn. They haven’t or can’t learn the new AI paradigm which appeared on in 2017.

That leaves us with three options: (1) Fire majority of them to make room (2) Demote the remaining good data scientists to sort of higher level AI analysts who conduct experiments and compare vendor/HuggingFace models. (3) Hire true data scientist who are either experienced with or educated on the current paradigm of AI, that is those laid off from big techs and those fresh graduates who learned current paradigm of AI at school.

Architects and Data Scientists, the attic where all the dusts collect, need a desperate cleansing. With these two roles straighten up, Fidelity 2.0 may start!

I’ve noticed a lot of posts from employees who are feeling anxious, especially those with longer tenure. But if you look at the RTO and data science groups, most of the people who ended up on PIPs were actually under 40.
When a manager isn’t technical, they often rely heavily on the more experienced people in the team. Those employees frequently position themselves as “mentors” or “coaches” to younger staff — sometimes as a way to stay relevant. The reality is that in the age of AI and rapidly evolving tools, not everyone keeps their skills current, and performance gaps can show up. In some cases, that wrong team dynamic ends up putting younger employees at a disadvantage.
A big part of the issue is XOM’s structural design. Opportunities aren’t distributed evenly, and we end up having some people doing tasks for years.

I’m seeing many resignations on LinkedIn following the selections.

I’m seeing teams recruiting externally for data scientists after layoffs.

I’m hearing of unqualified people, with no background in data science, being given positions, whilst others are let go.

I’m hearing that AI has been moved under CNE?!

WTAF?