Designing systems for resilience in an uncertain world

Digital systems today operate across interconnected components, global supply chains, and rapidly evolving AI/ML stacks, producing systemic risks that cannot be managed by perimeter controls alone. This presentation argues for a pragmatic return to engineering fundamentals: design for resilience—anticipation, withstanding faults, recovery, and adaptation—so organisations can survive and learn from inevitable compromise. Drawing on high-impact incidents from 2020–2025, practical frameworks, and measurable metrics, the talk provides an operational roadmap for embedding resilience into development, procurement, and operations.

Modern software and infrastructure are composed of layered dependencies, third-party services, and machine-assisted components whose provenance and behavior change rapidly. Supply-chain compromises, malicious or hijacked open-source packages, and insider vectors create nth-party blind spots that propagate risk across sectors. Geopolitical tensions and regulatory pressures further complicate trust in vendors and technology, forcing organisations to manage not only technical vulnerabilities but also legal and operational uncertainty.

We unpack forensic timelines and root-cause mappings from representative incidents between 2020 and 2025 to surface recurring failure modes and decision tradeoffs. Examples include ransomware-driven OT shutdowns, managed-service supply-chain compromises, and large-scale healthcare and telecom outages that illustrate how credential misuse, weak provenance controls, and inadequate segmentation produce cascading impacts. Each case is used to extract concrete engineering lessons: where design choices failed, which controls slowed attackers, and how recovery succeeded or stalled.

The talk defines four actionable principles and translates them into engineering practices and acceptance criteria:

Anticipation— threat-informed design, supplier risk modeling, and adversary emulation to reveal brittle dependencies.

Withstanding Faults — robust segmentation, least privilege, and fail-safe defaults that limit blast radius.

Recovery— immutable backups, integrity verification, and rehearsed playbooks enabling rapid, confident restoration.

Adaptation— post-incident learning loops, supplier remediation contracts, and continuous improvement baked into release cycles.

We present ten operational resilience metrics and show how to instrument them across CI/CD, SRE, and SOC workflows: MTTD, MTTR, RTO, RPO, exposure rate, phishing susceptibility, SBOM coverage, patch lead time, mean time to contain, and dependency depth. For each metric we provide: a concise definition, recommended data sources, target thresholds, and examples of how to enforce them as sprint KPIs and release gates. Practical guidance covers dashboard design, alerting strategies, and how to convert metric shortfalls into prioritized engineering work.

Frameworks such as NIST CSF, MITRE ATT&CK, and ISO are reframed as mapping tools for capability assessment and threat-informed defense rather than compliance endpoints. A highly recommended presentation for Executives, Board members, Senior Management, Fiduciary officials & Security Professionals to take a step back and assess how we can work together to infuse resilience into not only our systems, but our own thinking and prepare for the next era.