Nebius judges: start with Digital Wellness Academy.
SVTech Consulting Services LLC is the corporate applicant and the SoloFrame / Mono-PaaS home. Digital Wellness Academy is the healthcare product entered for the award.
If you arrived from the SVTech domain, this page is the handoff. The proof packet lives on DWA because that is where the MAIA safety architecture, clinical content system, data flywheel, and Nebius compute ask are documented.
Watch the DWA clinical workflow first.
Provider notes enter the portal, the system runs a privacy scan, recommends courses, sends the student a learning-path link, and later supports a student-owned progress share link.
SVTech is the company. DWA is the award entry.
The application points at SVTech because the LLC, platform architecture, and Mono-PaaS thesis live here. The evidence the committee needs first is the DWA technical packet.
SVTech Consulting Services LLC
The legal entity, founder story, and platform company behind SoloFrame / Mono-PaaS.
Digital Wellness Academy
The clinical education product with MAIA safety, provider workflows, 922 lessons, 21 assessments, and the data flywheel.
Nebius unlock
Production-class in-house coach LLM, faster MAIA retraining, multilingual coverage, and per-tenant fine-tunes.
DWA Nebius proof packet
The committee proof map: agentic AI shipped, MAIA safety, data flywheel, cost model, manifest platform, and Nebius unlock.
Open packet ->90-second technical walkthrough
A concise narrative through the safety architecture, provider loop, cost architecture, and compute ask.
Review ->SoloFrame / Mono-PaaS context
The corporate platform layer behind DWA: manifests, tenancy, routing, modular monolith discipline, and multi-vertical deployment.
Review ->Research context, shipped feature evidence, and award copy.
These materials are mirrored on SVTech so judges who start from the corporate application route can open the complete material directly without losing the DWA context.
Shipped feature brief: DWA platform intelligence
How the shipped RAG workflow intelligence layer extends into provider prep, clinical-notes-to-learning-path workflows, PHI isolation, and patient compliance reporting.
Open feature brief ->DWA Nebius award-page copy packet
Profile, innovation, traction, technical differentiation, Nebius infrastructure fit, and closing statement.
Open copy packet ->Education and AI reskilling research report
Market context tying Mono-PaaS to AI reskilling demand, vertical academies, and school-in-days deployment.
Open research report ->The next proof layer is clinical workflow intelligence.
This section summarizes the RAG, clinical-notes, and patient-compliance-report architecture: the same curriculum, RAG, MAIA, provider portal, and audit patterns extend into higher-value clinical workflows without turning the LLM into a clinician.
The curriculum becomes runtime knowledge.
DWA already grounds coaching and provider-facing search in its own evidence-linked content corpus. This section specifies the next layer: hybrid retrieval, reranking, and per-surface retrieval rules across coach, onboarding, provider prep, and clinical workflow surfaces.
Clinical notes point to learning paths.
The proposed notes-to-learning-path workflow keeps the clinician in control: de-identify the note, extract clinical themes, map those themes to DWA lessons through RAG, then present a draft path for provider review and sharing.
Between-session work becomes observable.
The patient compliance report spec turns completed lessons, assessments, thought records, tracking logs, and coaching interactions into a patient-controlled report scoped to a date range instead of a vague engagement claim.
Technical sophistication
The answer moves from "AI coach plus distress classifier" to a multi-surface intelligence system: curriculum-grounded RAG, MAIA gating, provider search, notes-to-learning-path drafts, and engagement reporting.
Real-world impact
The loop becomes visible: a clinician can assign evidence-linked between-session work, and the patient can bring back a dated record of lessons, assessments, thought records, logs, and coaching engagement.
Nebius infrastructure need
The compute story becomes sustained production workload, not a one-time training ask: MAIA retraining, coach-model work, continuous embedding, reranking, PHI-safe redaction/generation chains, and batch report generation.
DWA closes the loop between clinician guidance and patient practice: provider-led notes-to-learning-path drafts, patient-controlled engagement reports, and curriculum-grounded RAG running inside a PHI-safe multi-model architecture. Because compliance, safety gates, and RAG are baked into the engine, new AI capabilities can be composed from existing building blocks in hours instead of quarters.
This addendum does not claim autonomous treatment planning or diagnosis. The architecture keeps content knowledge in RAG so the curriculum can change without retraining, reserves fine-tuning for MAIA and future coach-model work, and leaves clinical decisions with the provider.
Platform globalization is proven. Clinical globalization is planned and gated.
DWA is English-first today because clinical localization is full-market healthcare work. It requires local clinician review, jurisdiction-specific crisis resources, consent language, and jurisdiction-specific compliance posture.
The platform underneath has already proven its first GTM-OS localizations through the LatAm (Spanish) and Brazil (Portuguese, just shipped) market surfaces: same core engine, localized buyer personas, market-specific workflows, regional messaging patterns, local integration choices, and no platform fork. That is a repeatable GTM-OS localization pattern, not the only geography. DWA follows the same architecture globally, with a higher clinical validation bar in each market.
SVTech does not claim that localized DWA markets are clinically ready today, that MAIA is validated across target languages, that jurisdiction-specific crisis workflows are implemented, or that jurisdiction-specific clinical compliance has shipped. The honest line is: localization without forking is proven at the platform layer; global clinical localization launches only after local review, crisis routing, consent, compliance, and model-validation gates clear.
Three things separate a platform from a Claude-shell-with-prompts.
Where AI is bounded out. How research enters the build. What the gate looks like that says no. Each item below maps to a published framework, not a vibe.
AI bounded OUT of the crisis path.
MAIA classifies signals. A deterministic state machine with 7 PMHNP-curated templates generates every crisis response. No LLM ever writes the safety reply. Excessive-agency risk is closed at the architecture layer, not patched at the prompt layer.
Promotion gate has already rejected models.
A model can clear in-distribution accuracy and still fail real-world spot-check. That has happened here. Receipts live in the repo. Strong benchmark numbers do not ship to users until the human gate signs off.
Research tiers gate every clinical claim.
Four-tier evidence hierarchy. Tier-1 (peer-reviewed) and Tier-2 (clinical guideline) are required to ground any shipped clinical or behavioral claim. Vendor blogs sit at Tier-4 and cannot back a production assertion. Research briefs expire on a clock.
Three-layer prompt-injection defense, CI-enforced.
Input sanitization, delimiter hardening, output validation. Tenant-leak harness in CI. PHI-aware logging redacts when compliance.phi: true. Per-conversation cost tracking with automatic model swap at $0.15.
Quarterly governance review against the CSA AI Security Maturity Model. SBOM + signed releases on every production artifact. AI model inventory + provider allowlist version-controlled in repo.
The technical packet answers the skeptical questions first.
The strongest evidence is not a pitch claim. It is shipped architecture, measured cost control, deterministic safety gates, and a product surface running on the same platform thesis described here.
1 founder. 922 lessons. 0 new hires.
DWA shows agentic AI used as an operating system for production output, not as a demo layer. The architecture scales the output. The org chart does not have to.
Compliance is a flag, not a fork.
DWA and GTM-OS run on the same SoloFrame / Mono-PaaS engine with different manifests: PHI-aware MAIA routing for DWA, noop classifier for GTM-OS.
$0.594 to $0.0955 per 20-turn session.
The six-layer cost architecture produces a 6.2x reduction today. Nebius compute unlocks Layer 6: the in-house coach LLM and per-tenant fine-tunes.
42 state-machine tests. Zero LLM calls in CI.
The MAIA promotion gate is deterministic enough to reject a model that misses the bar, with classifier-before-LLM routing and fail-closed behavior.
Recommended review path
For Nebius evaluation, start with the DWA proof map. Use SVTech for corporate context, SoloFrame / Mono-PaaS architecture, and the portfolio-wide platform thesis.
Go to the DWA Nebius proof map ->New on the DWA packet: a six-criteria scorecard and the global-impact case.