EBYSSLABS

AI systems engineering, governance visibility, and privacy-first systems. observe it or not.

I’m Ron Reed — an independent engineer focused on governance visibility, long-horizon system behavior, retrieval control points, and privacy-first software design.

I built RISWIS, a governance-first retrieval system that makes ranking logic visible before generation, CAS 2.0, a methodological evaluation system for observing drift, stability, and recovery in adaptive systems when intervention is deliberately withheld, and FARMSLite, a local-only lifestyle journaling app designed around reflection and presence — without accounts, tracking, or cloud dependencies.

Technical stack (selected): Python (SymPy, NumPy, Pandas, Matplotlib) · Dart / Flutter · reproducible logging, validation artifacts, and archival discipline.

Current work: governance-aware retrieval visibility, frozen-parameter long-horizon evaluation design, drift characterization, and auditability. Public artifacts are archived for transparency, review, and independent inspection.

S³P: Safety · Privacy · Presence. No hype. No performance claims. No persuasion.

RISWIS

Governance-first retrieval — ranking made visible before generation.

What it is
  • Retrieval Integrity & Structured Weighted Information System.
  • A retrieval framework that separates semantic similarity from governance weighting.
  • RISWIS acts as a Governance Retrieval Layer (GRL) between retrieval and generation.
  • Makes ranking behavior visible before an LLM or downstream layer ever sees the data.
  • Designed for auditability, traceability, and controlled observation of ranking shifts.
Why RISWIS
  • Most retrieval systems expose results but hide the ranking logic that shaped them.
  • Policy influence is often real, but not inspectable.
  • RISWIS makes it possible to see when policy overrides semantics and by how much.
  • GRL is the control point where ranking becomes an explicit decision rather than an implicit side effect.
Links
GitHub Substack

CAS 2.0

Long-horizon drift & stability — observed, not steered.

What it is
  • A methodological evaluation system for long-horizon observation of adaptive dynamics.
  • Frozen parameters by construction; no optimization, tuning, retraining, or corrective control.
  • Elapsed time is treated as the primary stressor to reveal second-order and long-tail behaviors.
  • Evaluated across 525+ structured configurations and 25.5+ million simulation steps.
Why CAS
  • Short runs often miss slow drift, partial recovery, and late-emerging degradation.
  • Adaptive “fixes” can mask intrinsic dynamics that only appear under sustained exposure.
  • Auditability and reproducibility are treated as first-class constraints.
Research artifacts
GitHub (CAS2 notes) Zenodo OSF ORCID

FARMSLite

Local-only lifestyle journaling — privacy-first by design.

What it is
  • An offline-first journaling app to record daily life inputs across Food, Activity, Rest & Reflection, Mood, and Sleep — stored locally.
  • No accounts. No tracking. No analytics. No Bluetooth dependency.
  • Designed as a quiet mirror for reflection, not a coaching or advisory system.
Why FARMSLite
  • People should be able to reflect without surveillance or extraction.
  • Data stays on-device by default; the user remains the owner and operator.
  • Not a medical, wellness, or advice product.
Links
Website Google Play

Contact

Follow, review, or reach out.

Links
X LinkedIn GitHub Website
Ethical stance (S³P)
  • Safety: observational framing; no operational control claims.
  • Privacy: local-first design principles; no behavioral manipulation.
  • Presence: systems built to support attention and reflection, not extract it.

Licensed under the Ebysslabs Ethical Use License (no military / surveillance / coercive use).