Page 577 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

3 Use Case Requirements

REQ-01: It is critical that the data pipeline is capable of ingesting at least 5 GB of data per day
while maintaining an end-to-dashboard latency of less than 5 minutes. This ensures real-time
visibility and responsiveness in data-driven decision-making. 4.6: Finance

REQ-02: It is expected that the risk engine API delivers risk scores with a response time under
500 milliseconds at the 95th percentile (P95). This level of performance is necessary to support
time-sensitive applications and user experiences.
REQ-03: It is expected that machine learning models are retrained on a weekly basis to maintain
relevance and accuracy. Additionally, it is critical that immediate retraining is triggered if data
drift is detected, as indicated by a Kolmogorov–Smirnov (KS) statistic greater than 0.2.

REQ-04: It is critical that all API endpoints comply with OpenAPI 3.0 standards and are secured
using OAuth 2.0 protocols to ensure data integrity, authentication, and secure access.

REQ-05: It is critical that IoT communication links are secure, encrypted, and achieve a minimum
uptime of 99%, thereby ensuring reliability and resilience in continuous operations.

4 Sequence Diagram

541