Doing More with Less
The tech industry in 2024 is under pressure to optimize resources. Technology and data leaders are asked to integrate more data to support new AI-driven features while simultaneously being forced to reduce costs and headcount. Judging by the recent layoffs at a.o. Google, Amazon, Meta, Twitch, Spotify, and Discord, even the largest tech companies are not immune to this trend toward increased efficiency.
The Impact of AI on Layoffs vs. Economic Factors
The growing capabilities of LLMs are reshaping the job market, and the data space is no exception. While it’s difficult to estimate to what extent AI progress has contributed to the growing waves of tech layoffs, many companies are cutting costs in their established lines of business and reallocating that budget toward AI development. Dropbox reduced its headcount by 16% last year and reallocated those resources toward hiring AI specialists in order to “stay competitive”.
Economic factors such as a slowdown in VC funding and some (late) post-pandemic adjustments also play a role in the headcount decisions.
Implications for Data Engineering
As organizations seek to do more with less, there’s a growing demand for generalists proficient in cloud-native technologies, data, AI, and platform engineering. This shift is steering the field away from highly specialized roles, such as ETL or BI engineers, in favor of a broader range of engineering skills. Data engineering teams in 2024 start resembling software engineering teams. This happens partially thanks to the growing maturity of data engineering as a discipline and partially out of necessity: data teams are expected to deliver more with less, and this requires building data products faster, often in smaller teams than before.
On the other hand, software engineers working on AI-driven features or data products start taking over many data engineering tasks, such as data cleaning, validation, and governance, because the quality of AI-based products depends on the quality of the underlying data. Tuning an LLM on bad data won’t lead to good outcomes for the business, regardless of how many GPUs we throw at it. You may notice that the boundaries between what software and data teams do are getting blurry in 2024.
To continue reading this article, click here.