Skip to content
View mgillr's full-sized avatar
11 followers · 71 following

Organizations

@Truedone

Block or report mgillr

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
mgillr/README.md

Ryan Gillespie

Complex integrated systems and capital delivery. Engineering background; building tools for deterministic distributed AI on the side.

Switzerland · data@optitransfer.ch

What I work on in my own time

crdt-merge — a merge library where every strategy converges: dataframes, JSON and model weights, merged without a coordinator. Python, TypeScript, Rust and Java editions. Paper: arXiv:2605.19373.

acfa-rs — Byzantine-robust federated aggregation with receipts anyone can re-execute. Open, Apache-2.0. Paper: arXiv:2607.10305.

Model merging — a long-running experiment in folding fine-tunes and whole architecture families into a single model, gated so it only keeps changes that hold up on measurement. Collectives published under Optitransfer on Hugging Face.

Rail — a research project on onboard train integrity for pure ETCS Level 3: what carries the safety argument once the track circuits go. Machine-checked where it can be, honestly approximate where it can't (yet). Patents GB 2607132.4 / GB 2608127.3 filed.

Pinned Loading

  1. crdt-merge crdt-merge Public

    Conflict-free merge for DataFrames, JSON, ML models & distributed agents — powered by CRDTs. The first merge library where every operation is mathematically guaranteed to converge.

    Python 7 1

  2. acfa-rs acfa-rs Public

    Train one machine-learning model across many parties who don't trust each other. A dishonest minority can't skew the result, and every party computes byte-for-byte identical output on any CPU or ar…

    Rust

  3. agentkey agentkey Public

    One signing key per AI agent role. Ed25519 keys bound to a person's SSH key, DSSE envelopes over in-toto statements, offline verification, exit codes a pipeline can trust. Zero dependencies, Node 22.

    JavaScript 1

  4. proofline proofline Public

    Independent V&V for coding agents. The agent says done; Proofline re-runs every check against a contract, rejects claims the evidence does not support, and gates the push. Zero dependencies, Node 2…

    JavaScript

  5. Truedone/truedone Truedone/truedone Public

    Checks whether a pull request's tests prove its change: red without it, green with it. Flags skipped, focused and weakened tests in pull requests from AI coding agents such as Claude Code, Copilot,…

    JavaScript

  6. sovereign-hospital-federation sovereign-hospital-federation Public

    Leaderless, coordinator-free federated learning across hospitals: patient data stays sovereign (GDPR/nFADP). Peer-to-peer CRDT gossip, Byzantine-robust ACFA aggregation with verifiable receipts, qu…

    Python