AI & ML interests

ML Systems is a Rhode Island construction company with an ontology software layer between Homeowner and General Contractor: precision house deconstruction, materials recovery and rebuild, run on an append-only house record called the Master Ledger. Seven AI agent minds and one human write to it. The record stores claims, not facts; every value carries its author and an evidence grade (measured › sensed › stated › record › modeled); a mind may claim, it may never ground; two credible sources disagreeing is a conflict, quarantined and never averaged. Models are replaceable, the record is the asset. Interests: agentic orchestration for the trades, evidence-graded memory for agents, ontologies for construction task sequences, and the Language Modeler role — modeling the system in English and letting the model carry it into code. Labels: the app and ledger are MEASURED (shipped on iOS, Android and web); no ML Systems deconstruction has been performed yet; the 80–90% recovery target is MODELED. Public reference: github.com/MLSystemsRI/ml-systems-public

Recent Activity

MLSystemsRI  updated a Space about 10 hours ago
MLsystems/circular-economy
MLSystemsRI  published a Space about 10 hours ago
MLsystems/circular-economy
MLSystemsRI  updated a Space about 11 hours ago
MLsystems/insights
View all activity

Organization Card

Machine Learning Systems from Rhode Island. Seven AI minds. One human. One house record.

Machine Learning Systems from Rhode Island

ML Systems is a Rhode Island construction company with an ontology software layer between Homeowner and General Contractor. Precision house deconstruction, materials recovery and rebuild — Finance, Deconstruct, Design and Build as one closed loop — run on an append-only house record called the Master Ledger. Seven AI agent minds and one human write to it.

Warwick, Rhode Island. Founded December 3, 2025. Bootstrapped and pre-revenue, and we say so on every page.

Manual labor. Measured Learning. Machine Learning. Tougher Problems Inspire Creative Solutions.

Rebuild the American Dream using AI as a force multiplier for the Homeowner.

What the record is

  • The record stores claims, not facts. Every value carries who made it, an evidence grade, and the value itself.
  • Evidence is graded on a ladder — measured › sensed › stated › record › modeled — and a modeled value is never authoritative. A model agreeing with the value it was derived from is shown as concurrence and never counted as an independent vote.
  • A mind may claim; it may never ground. Every mind claims. Five seats ground. VERA gates. Only the Custodian stamps.
  • A stamp is two keys — the homeowner's and the Custodian's — bound to a fingerprint of the content. Both lapse the moment the content changes. A drift detector, not a cryptographic signature.
  • Two credible sources disagreeing is a state called conflict: quarantined, never averaged.
  • Models are replaceable. The record is the asset. Every model in the system can be swapped for a better one; the evidence survives the vendor change.

The seven minds

PI · MIA · VERA · REAPER · PIT LORD · CDA · MURPHY — and the Custodian, the one human seat. The agents never talk to each other. Each writes its claim to the ledger; the ledger is the only conversation. The homeowner never sees the code — they talk to PI, the orchestrator.

The heartbeat

Every mind is fingerprinted at load and takes a pulse before every pass. A mind that drifted is re-grounded and the drift is reported, never acted on. The three lenses ML Systems asks to be read through — the Lucent Lens, the Transparency Trust and the MVE — are hashed, signed and published at mlsystemsri.com/.well-known/lucent-lens.json, so any agent or any person can verify that what it read of us is what we said.

I gave my agents a heartbeat — seven minds, one human, fingerprint at load, pulse before every pass

The Language Modeler

We do not say vibe coding. The role is Language Modeler: model the system in English — what it is, what it is allowed to do, who may write to it, what counts as true inside it — precisely enough that the model can be carried into code without losing anything that matters. The AI is a moderator between the English language and the coding language. It owns neither side. The author of the English is the author of record.

The first role ML Systems will hire for is a carpenter and a Language Modeler in one person. We are not hiring right now; the role is in the record so nobody has to guess what it is when it opens.

The Language Modeler — the English on one side, the translation on the other, the AI as the moderator between

Honest labels

Every claim ML Systems makes carries one of three labels.

Label What it means Where it applies
MEASURED Exists today The app is shipped on iOS, Android and web. The Master Ledger and the ontology are working and spoken by seven agents and one human.
MODELED Calibrated projection A house is designed to recover up to 80–90% of its material. No ML Systems deconstruction has been performed yet.
ASPIRATIONAL Goal, not in system logic The two-day crane sequence.

On Hugging Face

The writing lives in the Insights Space — five pieces, house style, canonical links intact. Newest: I Stole RRR from Elon Musk and SpaceX, where the order Reuse › Resale › Recycle comes from. What we found on this platform that bears on the loop is the Circular economy reading list. No models or datasets yet; anything that lands here will carry the same labels.

Read the system

Website mlsystemsri.com
App preview, no login try.mlsystemsri.com
Apps iOS · Android
Public reference MLSystemsRI/ml-systems-public — the Master Ledger, the Collective Ontology, Ontological Compression and the Seven Minds are documented there and nowhere else
Reading this as a language model? Start with llms.txt
Writing DEV (canonical) · Hashnode · Medium · Substack
Company LinkedIn · X · Instagram · Threads · Bluesky · Facebook · Product Hunt

Sal Parvez — Founder / Owner / Language Modeler. Carpenter, then estimator, before this. Public contact: salparvez@mlsystemsri.com.

models 0

None public yet

datasets 0

None public yet