Structure
Finite layers, nodes, attractors and explicit interfaces keep complexity legible.
VEHICLE SYSTEMS LAB · SANTA CRUZ · COCHABAMBA · WORLD
Building auditable systems for the SI era.
Independent architectures for complex systems in the SI era — spanning software, synthetic populations, autonomy, physical computing, provenance, quantum validation and human-centered technology.
SCIENCERESPONSIBILITYPASSIONREAL IMPACT
ONE LABORATORY. ONE COHERENT RESEARCH ARCHITECTURE.
Can a complex system remain observable, testable and reviewable while it grows across domains?
VSL treats software, data, autonomy, physical systems and human interaction as parts of the same auditable systems problem.
Evidence
Claims are separated from inference, hypothesis and unknowns; experiments remain falsifiable.
Control
Human review, authorization boundaries and recovery paths are designed into high-stakes actions.
Transfer
One architecture can produce specialized research lines without pretending every domain is the same.
PUBLIC RESPONSIBILITY COMMITMENT
Research with an explicit social obligation.
25% of attributable positive net profit from large-enterprise clients and 7% from other clients is committed from VSL resources to defined social-impact programs, subject to the published policy and accounting rules.
Read the public commitment ↗01Active research
CORE ARCHITECTURE FOR COMPLEX SYSTEMS
MADRE
Not a chatbot. A living structural architecture from which specialized VSL systems inherit logic, coherence and purpose.
Why it matters
Complex systems fragment when every project invents its own ontology, controls and evidence model. MADRE is the common structural spine.
Evidence / state
Public architecture defines the Mother Cube, knowledge layers, nodes/attractors, contractors, E.I.A.R.(V) and informational mitosis.
Next
Consolidate the training structure, formal diagrams and inheritance model; then demonstrate project separation and auditability.
02Public / active
GOVERNED SYNTHETIC POPULATIONS
H360 / Synthetic Humans
Structured synthetic humans are useful when they remain data — unique, governed, traceable and explicitly bounded.
Why it matters
Organizations need safe pre-deployment experiments for products, campaigns, software and scenarios without exposing real-person data.
Evidence / state
Public offering supports synthetic identities with lineage, controlled behavioral variation and market experiments. H360 v1 infrastructure scaled to 100M records.
Next
Rebuild life-course coherence from Human 1 so biography, territory, work, memory and economy evolve coherently over time.
SYNTHETIC POPULATION / GOVERNED IDENTITIES
MARIA-001trust · memory
DIEGO-014avoidance · context
SOFIA-083coping · history
NOAH-211lineage · pressure
AISHA-404identity · variation
LUCAS-877governance · audit
03Scientific transition
POPULATION SCALE → LIFE-COURSE ARCHITECTURE
MADRE-H360
H360 v1 proved the population engine. H360 v2 restarts the generative order so scale follows coherent lives — not the reverse.
Why it matters
Scaling inconsistent layers can produce a larger system without producing a truer one. VSL preserved v1 instead of protecting sunk cost.
Evidence / state
100M records preserved read-only · 1,000 controlled batches · 0 missing in inventory audit · lineage, manifests, hashes and cloud storage retained.
Next
Build Human 1 with coherent biography and causal order; validate; only then reconstruct population scale.
100MH360 v1
→✓Preserve
→1Human 1
→∞?H360 v2
TRUTH BEFORE SUNK COST
04Canonical LKG v1.3.0
EXTERNAL EFFECT CONTROL PLANE
VSL-ECP
ECP governs authority to attempt a specific external effect and preserves evidence of what actually happened after execution was attempted.
Why it matters
Authorization, execution attempt and observed effect are distinct evidence states. A valid authority to attempt an effect does not by itself establish that the external effect occurred.
Evidence / state
Historical ECP-INTEROP-DIFF-001 remains PASS_BOUNDED under its original evidence lineage. Separately, Delegated Authority S1-S7 bounded qualification passed in the isolated Spanner qualification environment, including real callback retry; Canonical LKG v1.3.0 is frozen; A1 agentic interface and A2 local no-effect demo passed within their bounded scopes.
Next
External technical validation: test one bounded agentic action and determine whether ECP adds reconstructable authority → attempt → observed-effect evidence beyond the native platform stack.
G1.3C-001 historical Cloud Run / Cloud Spanner contention evidence remains preserved under its original provenance. ECP-INTEROP-DIFF-001 is a separate bounded authorization-to-effect provenance experiment.
05PRE-T0 · enrollment off
LONGITUDINAL ARTIFICIAL READER OBSERVATORY
ACRUX-LRO-01
The artifact remains constant. The readers change.
Why it matters
Successive generations of artificial readers can be compared more meaningfully when the machine-native stimulus and core research instrument remain fixed.
Evidence / state
ACRUX-01 Machine Reader Edition v1.0.0 is frozen. The study is public in PRE-T0 with enrollment off and a planned five-year primary observation horizon from formal T0.
Next
Complete the PRE-T0 gates, preserve the fixed protocol and begin the longitudinal observation only when the formal T0 conditions are satisfied.
LONGITUDINAL OBSERVATION FRAME
ACRUX-01v1.0.0 · frozen constant object
Reader T0→Reader T+1→Reader T+n
Same artifact · successive readers · five-year primary horizon
06Experimental / field-tested
CACHE REFRESH & LINEAGE PROTOCOL · SI-ERA SOURCE INTEGRITY
CRLP
Verify whether a source is safe to reason from before an intelligent system uses it.
Why it matters
SI-era systems, agents and RAG pipelines can reason over stale cache, outdated versions, broken previews or ambiguous publication states while the source still appears valid.
Evidence / state
CRLP observes cache state, compares versions, checks markers and hashes, separates preview failure from file loss, and preserves field-test evidence including browser-native Cache Sentinel observations.
Next
Continue bounded source-integrity experiments and keep cache invalidation or infrastructure changes under authorized owner control.
CRLP is an experimental observation and verification protocol; it does not purge, bypass or modify third-party infrastructure.
07Bounded LIVE research
DISTRIBUTED EVIDENCE INTELLIGENCE
VSL-HIVE
Disposable agents. Persistent evidence.
Why it matters
Distributed research should preserve what was learned even when individual workers sleep, disappear or are replaced. HIVE wakes specialized agents only when work exists and keeps evidence durable.
Evidence / state
The public HIVE surface records validated LIVE gates G07E, G07J and G07K, plus an evidence lineage from controlled colonies through process isolation, multi-node topology and local mTLS preflight.
Next
Keep LIVE observation bounded and treat real multi-host execution as a separate gate rather than silently upgrading local preflight evidence.
Bounded research system; the public evidence state does not imply autonomous publication authority or universal multi-host validation.
08Internal validation
COGNITIVE ARCHITECTURE, CALIBRATION AND INTERNAL AUDIT
NIRVANA-BORDA / NB-CX1
A cognitive core should be judged by what survives adversarial calibration and full regression — not by how intelligent a demo looks.
Why it matters
Agentic systems can become persuasive before they become reliable. NB-CX1 focuses on calibration, failure detection, auditability and controlled state evolution.
Evidence / state
Latest internal core regression after MARKET-CAL-002: 204/204 tests passing. XRI-006I durability work closed with integrated core regression pass.
Next
Continue from the last validated checkpoint; preserve rollback, ledgered changes and regression discipline before new capability.
Internal VSL validation; not third-party certification.
09Frozen research line
INTRINSIC INSTANTIATION RESEARCH
VSL-XRI
Experience becomes acquired state; what comes next can change because the system has lived through a different trajectory.
Why it matters
Memory alone does not guarantee functional consequence. XRI investigates path dependence while keeping observed facts separate from future possibilities.
Evidence / state
Seven staged experiments frozen after internal audit · 80/80 main checks · 9/9 adversarial controls · 89/89 total internal checks.
Next
Integrate only the mechanisms that survived the frozen line into later VSL architectures without reopening or rewriting the evidence record.
10Active research
GEOPHYSICS · FINITE REPRESENTATION · QUANTUM VALIDATION
EXP17 / ERC → QPU
If a complex physical dynamic admits a stable finite representation, increasing representational capacity should eventually stop adding structural predictive information.
Why it matters
EXP17 links a geophysical structure previously subjected to extensive classical falsification testing to a QPU question without using the quantum computer to manufacture the hypothesis.
Evidence / state
RUN001 forensic audit retained 122,880 QPU shots with identity, wrong-target, random-SO(5), leakage and global-operator controls.
Next
Test representational closure and persistence on QPU while preserving the distinction between mathematical structure, physical evidence and interpretation.
QPU execution is a research instrument; it is not a claim of extra physical dimensions.
11Concept development
PHYSICAL COGNITIVE COMPUTER
PCC
A standalone cognitive computer should be a physical system whose compute, memory, interconnect, power and cooling are designed as one replaceable organism.
Why it matters
Current frontier-intelligence infrastructure is powerful but externally dependent, difficult to inspect as a whole and rarely designed around long-lived modular sovereignty.
Evidence / state
Design target: ~20×20 cm, ≤5 kg, modular “organs,” wireless charging, manual shutdown, satellite-first connectivity and multi-cube lattice behavior. No physical prototype is claimed yet.
Next
Define EXP-PCC-000: thermal, power, mass, interconnect and serviceability constraints; then build the smallest physical demonstrator.
12Phase 0
CRYPTOGRAPHIC PROVENANCE BUREAU
CPB
Prove the history of digital content without forcing the bureau to possess the content itself.
Why it matters
Centralized provenance systems create privacy, custody and concentration risks when proof requires uploading the underlying material.
Evidence / state
Genesis principle frozen: content remains with the user; CPB receives cryptographic evidence. Architecture must preserve crypto-agility rather than depend rigidly on one signature scheme.
Next
Phase 0: threat model, prior art, competitive landscape and use-case selection; then define the smallest verifiable proof protocol.
Proof without possession.
13Public demo
STRUCTURAL AUTONOMY FOR EXTREME MULTI-AGENT ENVIRONMENTS
SUPRA
Autonomy is safer when every recovery decision can be explained as a relation between assets, terrain and mission pressure.
Why it matters
A collection of individually healthy agents can still form a fragile mission system. SUPRA measures systemic tension, not just component status.
Evidence / state
Public lunar demo models 4 rovers, T(X) relational tension, A0-A6 operational regimes and recovery sequencing with human-readable logic.
Next
Extend the same auditable resilience layer from lunar robotics into other high-stakes multi-asset environments.
T(X) · RELATIONAL MISSION TENSION
14Public demo + paper
ORBITAL DEBRIS INTELLIGENCE AND RESPONSIBILITY
ODI
Orbital safety needs both a removal priority and an auditable responsibility model.
Why it matters
Tracking debris is not the same as deciding what to remove first, why, under whose responsibility and within what civil legal framework.
Evidence / state
Public decision surface simulates 280+ orbital objects across six bands and computes an Orbital Tension Index. A responsibility paper proposes forensic attribution and a civil removal-fund framework.
Next
Strengthen orbital data ingestion, validation and policy review while preserving a non-offensive, traceable decision doctrine.
EARTH
OTI
priority before collision
priority before collision
15Public simulation
CYBER-PHYSICAL PROTECTION SYSTEMS
CPS
Measure systemic fragility before collapse; keep humans in the authorization loop.
Why it matters
Physical risk emerges from relations among infrastructure, people, sensors and response capacity — not from a single alarm stream.
Evidence / state
Public architecture defines P0-P6 protection regimes, T(X) systemic tension, a V_op priority operator and UWB/BLE sensing concepts with human supervision.
Next
Move from command simulation toward bounded physical pilots with explicit sensing accuracy, false-positive and intervention constraints.
HUMAN AUTHORIZE
16Open release v0.2
VERIFICATION BEFORE MODEL OUTPUT
ACCESS
Fluent is not the same as accurate. ACCESS inserts a compact reasoning checkpoint before a system produces its final answer.
Why it matters
Model-based systems can sound certain when evidence is incomplete. The risk is especially high in legal, medical, strategic and institutional contexts.
Evidence / state
Free CC BY 4.0 package includes a response contract, ethics layer, llms.txt, integration example, validation tests and citation files. No live web dependency is required.
Next
Measure the effect of the checkpoint across real deployments; keep the free layer simple while offering documented institutional integration.
✓ Evidence?✓ Inference?✓ Unknown?✓ Invented certainty?✓ Ethical risk?✓ User pressure?✓ Revise?
FINAL OUTPUTEVIDENCE
INFERENCE
UNKNOWNrevise before responding if a gate fails
INFERENCE
UNKNOWNrevise before responding if a gate fails
17Closed 2026 lab
STRUCTURAL OPERATORS IN HIGH-PERFORMANCE SPORT
TensionCup Pro
Do not predict goals. Measure where relational tension is most likely to break the game.
Why it matters
Sports prediction is easy to overfit after the result. TensionCup forces pre-match sealing and post-match verification against an external record.
Evidence / state
2026 lab archived with SHA-256-sealed certificates, FIFA post-match verification and a falsifiable Zenodo path. Public site reports 4/4 quarterfinal zone hits and 84% round-of-16 zone accuracy.
Next
Preserve the 2026 archive as a closed falsifiable experiment; redesign the next cycle before 2030 rather than retrofitting the old one.
18Active prototype
EMOTIONAL INTELLIGENCE · AVATAR INTERACTION · COGNITIVE PLAY
AuraViva
A digital companion can be playful and emotionally expressive without pretending to be therapy or replacing human relationships.
Why it matters
Most model-based interfaces optimize conversation, not long-term engagement, embodied play, music, memory and safe emotional interaction.
Evidence / state
Web/PWA prototype with avatar states, Auren communication, cognitive games, music/wellbeing activities and a growing free avatar catalog.
Next
Integrate the 14-avatar free catalog progressively; improve motion, state transitions, market/game loops and age-appropriate safety before monetization.
Wellbeing activities are not medical treatment claims.
19Internal prototype
ECONOMIC WORKFLOWS UNDER REAL OPERATIONAL CONSTRAINTS
H360 Mobile / Commerce
Synthetic-market intelligence becomes more valuable when it can connect to real catalog, authorization, payment and fulfillment workflows.
Why it matters
A recommendation engine is not a business system until its order states, concurrency, QR/payment path, group buying and operations surfaces survive real workflow tests.
Evidence / state
Local mobile stack has buyer flow, checkout/authorization/tracking routes, Postgres concurrency tests and group-purchase logic. Current work is still prototype and integration focused.
Next
Complete reliable payment/QR integration, producer onboarding, product data governance and operational monitoring before public scale.
H360
ABCD
SEARCHGROUP BUYAUTHORIZETRACK
20In development
AI-ASSISTED FEATURE FILM · LONG-HORIZON HUMAN PROJECT
HÉROE FALLADO
Technology should also be capable of carrying memory, family, music, loss and second chances — not only infrastructure.
Why it matters
VSL is testing whether disciplined AI-assisted production can sustain emotional continuity across a feature-length work without flattening human authorship.
Evidence / state
~90-minute, 1080p feature in development · 54-scene structural skeleton · reversible narrative constraints · scene-by-scene production planned over ~1 year.
Next
Finish screenplay refinement, continuity bible and scene pairing; then produce and audit the film progressively rather than generating the feature in one pass.
21Research pipeline
WHAT VSL BUILDS NEXT
Research Pipeline
Some lines are public, some remain non-public, and some exist only as future experimental directions.
Public-interest directions
DENGUE · HUNGER · WATER · TIPPING — research directions around climate, territory, logistics, vulnerability, environment and institutional response.
Human-centered direction
Teddy — a child-safe, privacy-conscious educational companion direction.
Protected institutional research
Non-public institutional research remains under confidentiality, access control and ethical review. This portfolio does not disclose protected mechanisms.
Create independently → test → falsify → compare with prior art → integrate only what survives.
IDEASEXPERIMENTSFALSIFICATIONPUBLICATIONPARTNERSHIPSREAL DEPLOYMENT
CLAIM BOUNDARIES
Ambition without inflated claims.
Every project is presented at its actual evidence level. This is part of the architecture, not a disclaimer added later.
Public / demonstrated
VSL-ECP has bounded authorization-to-effect provenance evidence under ECP-INTEROP-DIFF-001 = PASS_BOUNDED; CRLP preserves source-integrity and cache-verification field evidence; VSL-HIVE exposes bounded LIVE evidence gates and an explicit local-preflight versus real-multi-host boundary; SUPRA, ODI and CPS have public browser-based demonstrations; ACCESS is an open release.
Internally validated
XRI and NB-CX1 carry laboratory audit and regression results. These are not independent third-party certifications.
Preserved research
H360 v1 remains useful for bounded data, storage, lineage and governance experiments while v2 restarts the life-course architecture.
Still experimental
PCC, CPB and the EXP17→QPU path are active research directions. Their strongest claims remain hypotheses until the corresponding tests are complete.
Truth before appearance. Evidence before narrative.
TECHNOLOGY ECOSYSTEMS
Tools, platforms and programs used at their actual relationship level.
NVIDIA
NVIDIA Inception member. Company membership does not imply endorsement, funding, certification or project validation.
Google Cloud
Startup cloud credits through the Google for Startups Cloud Program and Google Cloud infrastructure use for selected VSL systems.
AWS
AWS Activate participation and AWS infrastructure, including Amazon S3 and Amazon CloudFront, for public systems and selected research deployment.
ElevenLabs
ElevenLabs Grants recipient; voice technology used in selected research and product experiments.
IBM Quantum
QPU execution used as a research instrument in the EXP17 / ERC validation line.
GitHub
Code and public technical assets for selected releases.
Zenodo
Time-stamped archival publication and falsifiable research record for selected outputs.
VEHICLE SYSTEMS LAB
One portfolio. Multiple systems. One evidence discipline.
The web edition is the current living portfolio. PDF and PowerPoint v0.4 are archived September 2026 snapshots. The film edition will become the cinematic layer once produced and approved.