Agentic AI & GenAIConference50min
Your Monitoring Dashboard Is Already Too Late: Detecting Application Anomalies with Machine Learning
Traditional monitoring often detects failures only after predefined thresholds are breached. Yet incidents typically develop gradually, leaving subtle warning signs across logs, metrics, transaction patterns, and telemetry. Recognizing these signals earlier is essential to identify emerging problems and act before users experience significant performance degradation.
talk.summaryAiDisclaimer
Mohamed BahyaouiSociété Générale GSC Morocco
talkDetail.whenAndWhere
Thursday, November 5, 09:30-10:20
LILAS
Your dashboard is green. Your users are already feeling the slowdown. By the time the alert fires, the incident has been unfolding for minutes, sometimes hours.
Traditional monitoring is excellent at detecting known problems once they cross predefined thresholds. But production failures rarely appear out of nowhere. They leave behind subtle clues hidden in logs, metrics, transaction patterns, and operational telemetry. The challenge is recognizing those signals early enough to act.
Traditional monitoring is excellent at detecting known problems once they cross predefined thresholds. But production failures rarely appear out of nowhere. They leave behind subtle clues hidden in logs, metrics, transaction patterns, and operational telemetry. The challenge is recognizing those signals early enough to act.
Mohamed Bahyaoui
I’m a .NET Software Engineer at Société Générale GSC MA, working on financial risk management systems. My experience spans software development, system integration, and production support, with a focus on building reliable and maintainable applications. I work primarily with C# and the .NET ecosystem, and I’m particularly interested in software architecture, distributed systems, machine learning, and modern engineering practices.