Kenny Ademolu

Cyber architecture and engineering leader building trustworthy AI-native systems for identity, security, resilience, and agent evaluation.

I work at the intersection of cybersecurity architecture, platform engineering, deterministic simulation, and AI systems. This GitHub profile is my public proof-of-work space: open-source tools, benchmarks, field notes, and experiments for testing complex AI-enabled systems before they touch real infrastructure, sensitive data, customers, or money.

Current featured open-source project

SynthWorld

Deterministic synthetic identity worlds with adversarial evidence and ground-truth answer keys.

SynthWorld creates safely fictional, connected identity populations for evaluating privacy, PII extraction, entity resolution, relationship inference, exposure analysis, and risk-calibration systems.

Unlike row-oriented test-data generators, SynthWorld creates coherent identity worlds containing planted relationships, conflicting records, aliases, misspellings, search collisions, breach observations, and measurable ambiguity.

Product-visible observations are kept physically separate from evaluator-only truth, allowing systems to be tested and scored without leaking the expected answers into their inputs.

• Install: pip install idcognito-synthworld
• Repository: github.com/bluntmachetti/synthworld
• Frozen benchmarks: Hugging Face
• Licence: Apache 2.0
• Current benchmark families: exact-span extraction, entity resolution, relationship inference, exposure analysis, and risk calibration
• Next areas: AI-agent and non-human identity, enterprise IAM, and identity-governance evaluation
• Full direction: SynthWorld roadmap

SynthWorld is deliberately a focused identity and benchmark layer rather than a general-purpose simulation engine. It can provide deterministic identity truth to wider systems such as Arena, EADS, ZeroID, and Aftershock.

How the projects fit together

• SynthWorld provides connected synthetic identities, adversarial evidence, and hidden ground truth.
• Arena evaluates agent behaviour and organisational decision-making inside deterministic, queryable worlds.
• EADS models enterprise topology, dependencies, disruption, and business impact.
• Aftershock evaluates coordination structures and agent societies under repeatable scenarios.

Together, they explore how complex human, enterprise, and agent systems can be tested before they are trusted.

Other featured public work

Aftershock

Deterministic agent-society benchmark harness and live observatory.

Aftershock compares coordination patterns—including scripted, solo-agent, swarm, and structured-society approaches—under identical seeded scenarios.

Disaster response is the first scenario pack, but the underlying kernel is being developed as a disaster-agnostic, multi-model and multi-scenario evaluation environment.

• Public deployment: aftershock.redoubtlabs.dev
• Focus: agent-society benchmarking, coordination protocols, deterministic simulation, and cost/outcome evaluation
• Includes: typed decision contracts, model-backed runs, cost accounting, replayable run records, scenario packs, and a FastAPI/React observatory

What I am exploring

• Synthetic identity and privacy evaluation
• Agent identity, authority, and delegation
• AI-native security architecture
• Agent evaluation and deterministic simulation
• Operational resilience and enterprise digital twins
• Secure platform engineering
• Threat modelling and architecture decision intelligence
• Human and agent decision systems

Writing and field notes

I publish occasional field notes on simulation, agent societies, AI safety, identity, privacy, and operational resilience.

• Personal site — projects, essays, and public work
• Arena research notes — enterprise worlds, agent behaviour, and decision-making under organisational constraints
• Aftershock field notes — disaster-response simulation and agent benchmarking
• IncidentGym field notes — incident-response simulation and cyber exercises

Independent lab work

Some of my public experiments and research are developed through Redoubt Labs. Product, infrastructure, and commercial work generally remains private while I maintain a clean boundary between public research, personal experimentation, and enterprise work.

Background

I lead cyber architecture and engineering work at global enterprise scale, with a background spanning identity, Zero Trust, security architecture, platform engineering, operational resilience, and AI governance.

Earlier in my career, I built and operated security infrastructure across financial services, aerospace, critical transport infrastructure, aviation, and Olympic-scale live operations. I still prefer to remain close to the systems and build working implementations myself.

Contact

• Email: [[email protected]](mailto:[email protected])
• LinkedIn: linkedin.com/in/kehindeademolu
• GitHub: github.com/bluntmachetti
• Redoubt Labs research: blog.redoubtlabs.dev/arena

Collapse