Data Science Nigeria
Artificial Intelligence Bootcamp 2026

DSN Mart operates a chain of stores across Nigeria, from small corner shops to flagship hypermarkets, spread across major urban centers, state capitals, and smaller towns. Like any growing retail business, DSN Mart wants to understand what drives sales at the product level across its different store formats and locations, so it can plan stock, pricing, and store investment more intelligently.

You’ve been given a snapshot of historical product-store sales records and asked to build a model that predicts total sales for a given product at a given store, based on what’s known about the product and the outlet it’s sold in.

This challenge is being used as the qualifying hackathon for the DSN AI Bootcamp. Your rank on the leaderboard is one of the inputs used to select participants for the bootcamp.

View challenge on Kaggle

You have joined a retail business as a Data Analyst. Management has already received a dashboard and a set of conclusions based on the supplied dataset. Your task is to independently review the data, determine what can and cannot be trusted, correct the analysis, and produce a more reliable dashboard and set of business conclusions.

What you are expected to do:

Review the supplied raw dataset and the existing management dashboard.

Identify data-quality issues that could affect the reliability of the analysis.

Clean and validate the data using appropriate analytical judgement.

Audit the management claims and determine whether each is correct, incorrect, misleading, or cannot be determined.

Build a corrected dashboard using either Power BI or Excel.

Compare the original dashboard with your corrected dashboard and answer the final analysis questions.

Submit a snapshot of your corrected dashboard together with your completed Answer & Submission Sheet.

Create an original GIS map that explores a real-world issue related to health, environment, or climate change using freely available datasets.

Requirements:

Use open datasets such as OpenStreetMap, Google Earth Engine, or national open-data portals.

Any GIS tool may be used.

The map must be your original work.

Goal: Build and evaluate LLM-based approaches for generating topic labels from news text in low-resource Nigerian languages, comparing classical baselines, zero/few-shot prompting, and fine-tuning, and publish the result as a Kaggle notebook.

One-line pitch: Most topic classification/generation work is English-only. This project benchmarks how well modern LLMs handle topic generation in four major Nigerian languages, and shows how much fine-tuning closes the gap.

Identify the problem. Showcase your AI/ML solution. Demonstrate the impact.

The project must be completed before the abstract is submitted.