-
151
Projects submitted
-
109
Published projects
-
47
Comments left
-
66
Number of votes
-
52
days until voting ends
Use of AI and Machine Learning in a Cybersecurity Framework
-
Project manager on customer side
Alexandro Brito
Grupo Nichos
CIO
-
Category
-
Nomination
-
Goals
The critical problem to solve is the vulnerability to advanced and stealthy cyberattacks, as our defenses are currently primarily reactive.
With this project, we aim to shift this paradigm toward a proactive and predictive model. Instead of merely reacting, the goal is for the AI system to learn baseline "normal" activity to detect anomalies and suspicious behavior in real time, even if they represent previously unidentified zero-day attacks, stopping them before they cause harm.
-
Timeline
december, 2025 — august, 2026 -
Project completion year
2026
-
Project scale
3497 human-hours -
Results
Drastic Risk Reduction and Early Detection:
The most impactful outcome is shifting from reactive defense to predictive prevention, enabling the AI engine to detect threats that traditional controls bypass.
Mitigated Incidents: During initial operating steps, the system is expected to proactively mitigate a substantial volume of security incidents.
Operational Efficiency and Cost Savings:
Mean Time to Detect (MTTD): Target reduction of at least 50%, spotting anomalies in minutes rather than hours.
Mean Time to Respond (MTTR): Target reduction of at least 40%.
Detection Precision: Slashing the false positive rate by at least 80%, substantially freeing analyst bandwidth for high-value tasks.
Return on Investment (ROI): Projected ROI expected within 12 months.
Organizational Culture Impact: The cultural impact is anticipated to extend beyond technical teams, shifting the overall corporate perception of security toward a data-driven, risk-aware mindset.
-
Project uniqueness
In a market saturated with promises of artificial intelligence, The real gap typically lies between conceptual theory and corporate reality. Many organizations announce machine learning initiatives that never make it past the proof-of-concept stage or limit themselves to integrating generic commercial tools where AI acts as a mere marketing buzzword. What makes this case unique is that it breaks away from theoretical inertia to confront the friction of real-world operations.
-
Software used
1. Software & Cloud Platforms
A. Data Pipeline and Machine Learning Core (GCP)
Google Cloud BigQuery
Vertex AI Platform:
Cloud DLP (Sensitive Data Protection)
B. Telemetry Ingestion & Endpoint Loggers
Log Shipper Agents:
Syslog Forwarders & API Gateways
2. Hardware, Infraestructure
Workstations for Analysts & Data Engineers
Dedicated Secure Network Connectivity
3. Support Systems, Operations & Governance
Identity and Access Management (IAM)
Operational Runbooks & Knowledge Base
-
Solution from the Global CIO catalog
The project does not use solutions from the Global CIO catalog
-
Project implementation complexity
1. Data Heterogeneity: Ingesting and standardizing diverse logs (data bases, firewalls, apps) via streaming ETL without schema breaks.
2. Threat vs. Noise Balance: Baseline modeling to slash false positives by 80% without overlooking stealth zero-day exploits.
3. Latency & Cloud Costs: Ensuring sub-minute detection while managing GCP BigQuery and Vertex AI workloads to avoid OpEx overruns.
4. MLOps & Model Drift: Mitigating performance decay and labeling errors in automated retraining pipelines with zero-downtime deployment. 5. Compliance & Explainability: Enforcing pseudonymization under GDPR/ISO 27001, auditable identity tracing, and explainable AI to eliminate black-box alerts.
6. Cultural Shift: Upskilling analysts from static SIEM rules to probabilistic ML scoring while maintaining cross-functional governance.
-
Description
Our company faces increasing exposure to sophisticated cyber threats that are not always detected in time by our traditional mechanisms, jeopardizing the integrity of sensitive data and operational continuity. In an industry such as pharmaceuticals—the market where Grupo Nichos operates—intellectual property and operational data are extremely valuable, and a successful attack could be devastating.
The critical problem to solve is the vulnerability to advanced and stealthy cyberattacks, as our defenses are currently primarily reactive.
With this project, we aim to shift this paradigm toward a proactive and predictive model. Instead of merely reacting, the goal is for the AI system to learn baseline "normal" activity to detect anomalies and suspicious behavior in real time even, if they represent previously unidentified zero-day attacks, stopping them before they cause harm.
The project is a proactive cybersecurity solution using Google Cloud, Vertex AI, and BigQuery to detect stealth threats, automate MLOps, cut false positives by 80%, and safeguard IP under GDPR and ISO 27001 standards.
-
Project geography
The project involves managing information generated by the majority of the organization’s systems and infrastructure, which operate in Mexico.
Grupo Nichos is headquartered in Mexico City, with additional offices in other Mexican cities (Guadalajara, Monterrey, and Cuernavaca), as well as in Guatemala and Spain.
Its primary systems and applications are hosted on the Google Cloud Platform (GCP); the organization also maintains on-premise systems located at its headquarters.
Once the project is established and yields the expected medium-term results, its scalable architecture will allow for an expansion in scope—incorporating applications, systems, and infrastructure that support operations in other Latin American countries and Spain.
-
Customer
Grupo Nichos
Only authorized users can comment.
We suggest you log in or register.