Skip to content
View anulum's full-sized avatar
💭
Im publishing projects derived from "God of the Math" in here
💭
Im publishing projects derived from "God of the Math" in here

Highlights

  • Pro

Block or report anulum

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.

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
anulum/README.md

Miroslav Šotek — evidence, computation, control

Website · ORCID · Sponsors · protoscience@anulum.li

Miroslav Šotek

Independent researcher and systems engineer at the Anulum Institute in Switzerland.

I build evidence-governed infrastructure for AI systems and multi-agent engineering: coordination with receipts, reliability guards for model output, and research stacks that reach from mathematics into executable hardware paths.

Claims are only as good as the measurements, artefacts, or verification that back them.

Stack: Python · Rust · Verilog · formal and process tooling where needed.

Start here

If you need…Go to
Parallel coding agents that do not clobber each otherSynapse Channel · docs
LLM claim guarding / factual-consistency checksDirector-AI · docs
Repository audit and remediation planningRigor Foundry · docs
Neuromorphic / stochastic computing researchSC-NeuroCore · docs
Coupled-oscillator / quantum simulation researchSCPN Quantum Control

Project documentation also lives under each repository’s Pages site and on anulum.li.

Lab map

Anulum is a lab stack, not a single product:

Director-AI reliability of model output
Rigor Foundry evidence-bound audit and remediation
Synapse Channel multi-agent coordination, claims, receipts
SC-NeuroCore neuromorphic / SC compute (Python · Rust · RTL)
SCPN suite control, plasma, phase, and quantum research paths

Typical stack use

  1. Rigor Foundry — find what is broken, unproven, or unsafe to claim.
  2. Director-AI — guard model output that will be trusted.
  3. Synapse Channel — run multi-agent work with claims, mailboxes, and receipts.
  4. SC-NeuroCore / SCPN — when the problem is neuromorphic, physical, or control-grade compute.

Research, validation, and product readiness are kept separate. Active development is not a readiness claim.

The pinned repositories on this profile match the selected-work table below. Other public repos are either SCPN-family research or supporting tooling.

Selected work

ProjectJobMaturity
Synapse ChannelLocal-first control plane for coding-agent fleets: claims, roles, durable mailboxes, receipts, audit, and federationUsable now — functional core, active development
Rigor FoundryEvidence-bound repository auditing and remediation planningUsable now — active hardening
Director-AIReal-time LLM guardrails: NLI + RAG fact-checking with optional claim-level streaming haltResearch active — functional system under validation
SC-NeuroCorePolyglot stochastic and neuromorphic framework (Python, Rust SIMD, Verilog, HDC/VSA)Research active — platform under continuous development
SCPN Quantum ControlEvidence-governed quantum simulation of coupled-oscillator synchronisationExperimental — preregistered research programme

Related control and fusion research lives in the SCPN suite (control, fusion-core, phase orchestrator, MIF-core).

Maturity labels

LabelMeaning
Usable nowInstallable, documented, CI-backed; still evolving
Research activeReal code and ongoing science; not a stability promise
ExperimentalExploratory; do not treat interfaces or claims as fixed
Evidence-boundPublic claims are tied to measurements or artefacts

Evidence, not slogans

Negative and null results are published when they are real. Public claims stay tied to artefacts: measurements, preregistered protocols, raw packs, or executable verification — not slogans.

Example: preregistered quantum-control protocols and hash-bound result packs in scpn-quantum-control.

Working principles

  • Evidence before claims.
  • Reproducible artefacts before presentation.
  • Clear boundaries between research, validation, and product readiness.
  • Cross-language implementations where performance or hardware integration makes them useful.
  • Honest failure records: negative results are part of the research output.

Collaboration

I welcome technically grounded collaboration in neuromorphic systems, reliable AI infrastructure, scientific computing, formal verification, and control.

A useful first message includes: problem, constraints, relevant prior art, and what evidence would count as success. Prefer email (protoscience@anulum.li) or the contact path on anulum.li.

I respond to technical proposals. I do not take on ungrounded hype work, “demo-only” science theatre, or claims that cannot be checked.

For sustained open work, GitHub Sponsors funds CI runners, quantum and hardware experiment time, and public docs — not marketing.

Transparency: These repositories span research software, developer tools, and product candidates. Active development does not imply production readiness or scientific validation unless a project provides explicit evidence for it.

I AM THAT

Anulum

Pinned Loading

  1. synapse-channelsynapse-channelPublic

    Neutral control plane for coding-agent fleets: claims, roles, mailbox reliability, receipts, audit, federation, dead-letter visibility, sandbox receipts, and cross-agent coordination. This is still…

    Python 4

  2. director-aidirector-aiPublic

    Real-time LLM hallucination guardrail — NLI + RAG fact-checking with opt-in claim-level streaming contradiction halt. Drop-in for any LLM backend.

    Python 2

  3. rigor-foundryrigor-foundryPublic

    Evidence-bound repository auditing and remediation planning.

    Python 3

  4. sc-neurocoresc-neurocorePublic

    Universal Stochastic Computing Framework for Neuromorphic Hardware — Rust SIMD engine, Python simulation, Verilog RTL, HDC/VSA, SCPN integration

    Python 13

  5. scpn-quantum-controlscpn-quantum-controlPublic

    Evidence-governed quantum simulation of coupled-oscillator synchronisation — Kuramoto–XY workloads on Superconducting Computers, raw counts + hash-bound result packs, preregistered protocols.

    Python 3

  6. scpn-phase-orchestratorscpn-phase-orchestratorPublic

    Evidence-first synchronization analysis and review-only control proposals for coupled rhythmic systems (Kuramoto/UPDE), with reproducible audits.

    Python 2