Distinguished Security Engineer — AI/ML Confidential / Trusted Compute • Detection & Response • Integrity-first telemetry • Zero-Knowledge Multiple Party Compute • Zero Trust
Cypherpunk pragmatist: I turn messy operational reality into resilient operational excellence.
Creator note: I stream incident response labs + resilient comms builds.
- Ship detection & response systems that hold up under real adversaries, real outages, and real constraints, keeping up with the agentic-enabled threats.
- Build tamper-evident / privacy-preserving telemetry for regulated, high-sensitivity environments.
- Secure AI/ML compute pipelines using strong isolation boundaries, policy enforcement, and verifiable integrity while saving $$$$$$.
- Reduced alert MTTR by ~78% by tightening detection quality, test coverage, routing, and on-call runbooks—paired with Zero Knowledge guardrails (identity-aware access, scoped credentials, and continuous verification) across global-scale, multi-cloud environments.
- Led incident response for events impacting 1,000s–1,000,000s of services and 10K–90M+ users (severity triage → containment → postmortem-driven hardening), including compromised-credential and trust-boundary failure modes typical of Zero Trust threat models.
- Built detections covering 20,000+ tactics/behaviors across endpoint + on-prem + identity + cloud telemetry, with measurable precision/recall tuning—plus privacy-preserving correlation approaches (tokenization, secure aggregation, and ZK/MPC-inspired analytics) to enable cross-domain insights without raw-data exposure.
- Building: MPC/ZK integrity primitives, IR workflows, emergency ops comms
- Learning: Morphogenesis & developmental dynamics
- Exploring: non-terrestrial edge compute (NTN / satellite) + constrained environments
- zk-Autograd — verifiable / integrity-minded zero-knowledge multi-party compute for ML training experiments
- Mimir — security research + systems thinking, packaged as a real repo
- GhostTelemetry — tamper-evident telemetry pipelines & trust primitives
- persuadable-defender — adversarial/behavioral security experiments with assured agentic alignment
- InfoSec Blueprints — practical, copy/paste-ready security playbooks & blueprints
- Ham radio / maker (W8MEJ): resilient comms, RF + networks + operational reality
- Streaming builds & ops experiments:
- Twitch: twitch.tv/w8mej
- YouTube: youtube.com/@w8mej
- Podcast: Zero Noise Collective
- Email: coding@haxx.ninja
- Identity / proofs: https://keyoxide.org/sephiroth@haxx.ninja
- LinkedIn: https://www.linkedin.com/in/w8mej
- X (Twitter): https://x.com/w8mej
- Bluesky: https://bsky.app/profile/w8mej.bsky.social
- Mastodon: https://mastodon.social/@w8mej
- Substack: https://w8mej.substack.com