Knowledgebase

Machine Learning Work in the Nigerian Context Print

  • machinelearningengineering, machine, nigeria, troubleshooting, billing, guide, howto, solution
  • 0

Local realities.

WHAT THE OPPORTUNITIES ARE

Problems with local data nobody else has addressed Fintech, agriculture, logistics, health and education applications Language technology for local languages Remote work for international employers

WHAT THE CONSTRAINTS ARE

Hardware cost and availability Cloud billing in foreign currency Connectivity for large data transfer Power stability for long training runs Fewer local senior practitioners to learn from

WHAT THAT ARGUES FOR

Transfer learning over training from scratch Smaller models Careful cost management Cloud capacity rented rather than hardware purchased

WHAT DATA CHALLENGES ARE DISTINCTIVE

Less digitised historical data Inconsistent records Address and identity data that is difficult to standardise Languages and varieties poorly represented in models

WHAT THAT LAST POINT MEANS

Systems must be evaluated locally, and frequently adapted.

WHY THAT IS ALSO AN OPPORTUNITY

Work on under-represented languages and local problems is genuinely novel.

WHAT TO BUILD

Things solving problems here, with data from here.

WHAT TO JOIN

Local practitioner communities, which exist and are growing.

WHAT TO BE REALISTIC ABOUT

That most local demand is for applied work, not research.


Was this answer helpful?
Back

Are you happy with your experience? Leave us a review on Trustpilot.


Trustpilot