Skills application
Building secure ML systems
In this skills application, you will apply what you’ve learned about secure infrastructure, ML-specific threats, team protocols, and secure workflow design.
Your goal is to design a more secure ML system for a healthcare company facing real-world security concerns.

Context
You are part of the ML team at HealthPredict, a healthcare analytics company developing a predictive model to forecast hospital readmissions. The model is trained on sensitive patient data and deployed via a public-facing API that integrates with hospital systems.
Recently, your team has noticed suspicious access patterns, raising concerns about:
- Model inversion attacks (where attackers try to reconstruct sensitive inputs from model outputs).
- Unauthorised API access.
You’ve been asked to assess the system’s security posture and recommend improvements to protect patient data, reduce vulnerabilities, and support secure ML operations.
Success criteria
- Identify three relevant vulnerabilities in the current system.
- Propose a secure infrastructure setup andworkflow improvements.
- Design two team protocols (Incident Response & Continuous Monitoring) with clear roles.
- Reflect on the most critical issues and the impact of recommendations on team culture.
Instructions
1. Vulnerability analysis
- Identify three key vulnerabilities based on the model’s lifecycle (data ingestion, training, deployment).
- For each, map the issue to a relevant security principle (confidentiality, integrity, availability).
2. Secure infrastructure and workflow design
- Recommend an infrastructure setup (Local, Cloud, or Hybrid) and justify your choice.
- Propose at least three workflow improvements (e.g., encryption, API security, access control).
3. Team protocol development
- Draft an incident response protocol for breaches.
- Draft a continuous monitoring protocol to proactively detect threats.
- Assign responsibilities to roles (Data Engineer, ML Developer, Platform Engineer).
4. Reflection
- Which vulnerabilities were most critical and why?
- How did your changes reduce risks?
- How could team culture influence the success of these protocols?
Type your analysis here...
Type your recommendations here...
Type your protocols here...
Type your reflection here...