Traditional education treats data science like a theoretical discipline. Students learn math, algorithms and coding in sanitised environments. But in the real world, data is dirty, stakeholders are impatient and models break in production at 2:00 AM. Employers, terrified of this gap, respond by inflating job requirements asking for five years of experience for roles that didn’t exist five years ago or knowledge in software or programme that will become obsolete in few months. They automate resume filtering, rejecting anyone without “Spark” or “Airflow” on their CV. The result? Graduates can’t get their first break, companies can’t find “ready-to-go” talent. At the end, both sides lose. Companies are counting their profits but not people they have impacted, and they have the audacity to analyse poverty level in their countries or across the globe. We’re training people for a party that has a very small guest list and blaming the guests for not being dressed well enough.

Here’s the audacious proposal: Treat the first year of a data career like a medical residency or a trade apprenticeship. Not another internship. Not a shadowing program. A paid, productive, project-based apprenticeship where juniors work alongside seniors on real production systems with explicit learning outcomes and failure tolerance.

Year 1 - Pilot:

Governments and tech consortia fund 50,000 apprenticeships across ten industries. Each apprentice works four days a week on real business problems, one day on structured upskilling. Employers get subsidised talent, apprentices get portfolio-grade deployments.

Year 2 - Scale:

Based on pilot data, expand to 200,000 apprenticeships. Partner with under-digitalised sectors (agriculture, supply chain, public health). These industries don’t need cutting-edge AI, they need basic predictive maintenance, demand forecasting and anomaly detection. Perfect training ground.

Year 3 - Normalise:

Make apprenticeship the default entry path. Remove “years of experience” filters for junior roles. Replace 5-round interviews with 2-week paid sprints where candidates solve real business problems under mentorship.

Why this works. First, it de-risks hiring. Employers don’t have to bet on a resume, they can evaluate performance in context. Second, it builds translators not just coders. Apprentices learn to ask the right business questions, communicate with non-technical teams and handle production failures. Third, it redistributes opportunity. Apprenticeships can be localised, remote or hybrid bringing talent to regions that have been ignored by the big tech hubs.

The hard truth, we cannot train our way out of this crisis. More MOOCs, more bootcamps and more certifications will only flood the market further. What we need is structured, supervised, paid practice: the kind that builds judgment, not just knowledge. The talent is already there. A million trained professionals are ready. What’s missing is the bridge.

Let’s build it not in a decade but in three years. Because the data economy can’t wait. And neither can they.

Sources:

Underlying data comes from the National Center for Education Statistics (NCES) through its Integrated Postsecondary Education Data System (IPEDS). Their reports tracking undergraduate degree fields show over 134,000 yearly bachelor's degrees conferred strictly in "Computer and Information Sciences and Support Services," a pool heavily augmented by tens of thousands of Masters and PhD level completions.

The primary benchmark is the National Association of Software and Service Companies (NASSCOM). Their landmark report on the State of Data Science & AI Skills in India and their updated studies on the State of AI-Native Talent outline that India possesses the world's second-largest AI/ML/Big Data analytics talent pool, fed by an ecosystem that graduates roughly 1.5 million technology and engineering students annually.

The baseline is derived from the European Commission's statistical office, Eurostat. Their annual ICT education and statistical overview datasets and tertiary graduation datasets track the steady influx of computing, mathematics and engineering graduates feeding an active workforce of over 10 million ICT specialists.

Data originates from the Ministry of Education (MOE) of the People's Republic of China. Official updates released by the ministry indicate that China's higher education system produced over 55 million university graduates across a five-year period, with a strong structural emphasis prioritising Artificial Intelligence, quantum technology and broad STEM disciplines.

The baseline originates from historical data compiled for the UK government's Data Skills Taskforce, which utilises data from the Higher Education Statistics Agency (HESA) to isolate the maximum potential pool of pure data science graduates coming out of UK universities.

Casey ‘Gbenga Adeleye writes from the Institution of Data Scientists and Analysts.