Data Innovation Hackathon

Track one: Innovative data collection from non-traditional sources
Challenge name: Smart observatory for measuring platform-based employment
Challenge statement
Design and develop an automated observatory for the digital labor market that estimates the size and characteristics of the workforce in the Gig economy. By leveraging publicly available digital traces. The proposed solution should address the “blind spot” in traditional administrative records, which often fail to capture workers operating through digital platforms adequately.
Background
The labor market is undergoing a structural shift toward the digital economy (ride-hailing applications, delivery platforms, and freelance work). However, a substantial proportion of individuals engaged in these activities are not registered within formal social insurance systems, creating a significant statistical gap that constrains an accurate assessment of the true scale and characteristics of the flexible workforce.
Objective
Develop a proof-of-concept (POC) for an intelligent observatory capable of estimating the number of platform-based workers and classifying them by activity type. The solution should generate an interactive dashboard that compares official statistics with digitally observed labor market indicators, providing evidence-based insights to support decision-makers.
Data sources
Integrate data from freelance platforms, digital search trend indicators, and app store analytics—such as the volume and frequency of user reviews—as proxy indicators for estimating the size and activity levels of the workforce. These sources are provided for illustrative purposes only. Participants are encouraged to identify, innovate, and incorporate any additional publicly available data sources that strengthen the robustness of their proposed solution.
Proposed methodology
Participants are encouraged to leverage open-source solutions and low-code/no-code platforms. The proposed approach should incorporate automated tools for extracting publicly available data, data integration frameworks for consolidating heterogeneous datasets, and analytical models designed to estimate patterns and forecast future trends.

Track two: Intelligent processing and AI-powered automated classification
Challenge name: Intelligent logical error detection and correction system (Semantic Guardian)
Challenge statement
Develop an intelligent verification assistant that integrates seamlessly with survey platforms to detect semantic inconsistencies and complex logical errors in real time (such as mismatches between reported job titles and educational qualifications). The solution should move beyond rigid, rule-based validation checks by incorporating adaptive, intelligence-driven mechanisms that enhance data accuracy at the point of collection.
Background
Field data collection processes often encounter logical inconsistencies and conflicting responses that compromise data quality and necessitate extensive manual post-collection processing. The situation underscores the need for an intelligent solution capable of detecting and resolving such issues in real time during the data collection process.
Objective
Develop an integrated proof of concept (POC) that connects a large language model (LLM) to the data collection system to evaluate input quality and assign each response a confidence score. The solution must demonstrate real-time interactive integration with the survey instrument and the capability to process data dynamically during collection—not merely present a standalone (fine-tuned) language model.
Data sources
After the screening phase, qualified teams will be provided with a file containing 50 to 100 anonymized records or survey forms that include common logical errors and contextual inconsistencies observed in real field operations. This dataset will be used to test the system’s ability to process cross-variable data and ensure the plausibility of the results.
Proposed methodology
Use Application Programming Interfaces (APIs) for Large Language Models (LLMs) in conjunction with few-shot prompting techniques and employ orchestration frameworks to ensure actual integration with the survey platform. The solution should include a live demonstration interface that demonstrates the detection of errors immediately upon data entry.
 

Competition Framework 

The competition adopts an “open participation” model at the national level:

• Operating model: A fully virtual (online) Hackathon to enable participation from all regions of Saudi Arabia, with the closing ceremony streamed through approved digital channels.
• Accepted Tools: Traditional programming languages (Python, R) and low-code development platforms, such as KNIME.