DiSabatino LLC

Core Competencies

• Governed LLM gateway architecture Multi-provider routing, API-credential lifecycle, per-key budget enforcement, throughput control, and full usage audit (NIST AI RMF Govern/Manage; NIST 800-53 AC/AU/IA families).
• AI model inventory & AIBOM discipline Model/version pinning; commercial-model drift treated as a first-class governance event that triggers re-review.
• FinOps for AI Consumption and cost governance as a controlled, audited axis, not an afterthought.
• Deterministic enforcement A patent-pending cascade architecture that places a deterministic rules engine in authority over LLM output, with human-in-the-loop guardrails on any rights-impacting decision. An LLM is never the final authority.
• Language-invariant agent harness Native-language interaction over a single-audit deterministic rule layer: one audit covers every language. Rules are enforced once in a language-neutral protocol channel; the model converses in the user's language. Supports Title VI of the Civil Rights Act of 1964 and the Section 1557 meaningful-access obligation for veteran-facing services, both of which rest on statute and remain in full effect.
• Security & compliance operations at scale EDR fleet deployment, enterprise MFA, full-lifecycle vulnerability management, and cloud-credential governance mapped to NIST 800-53.

Acquisition Policy Alignment

• OMB M-25-22 (AI acquisition; solicitations issued on or after September 30, 2025) The memorandum's obligations are architectural, and the gateway enforces them: multi-provider routing against vendor lock-in, a no-train posture on government data evidenced by the usage audit trail, licensing and data boundaries kept structural by running on customer-operated infrastructure, and the components needed to operate and monitor the system remaining with the agency.
• OMB M-26-04 (LLM transparency; applies to any LLM an agency procures) The Minimum Threshold for LLM Transparency asks vendors for governance tools additive to base prompts, model evaluation across models, and visibility into output provenance. Those controls are the product: a versioned control plane, a calibrated multi-provider evaluation harness, and a dual-channel protocol carrying machine-readable state alongside every response.
• NIST AI RMF and NIST SP 800-53 The memoranda are what a contracting officer must satisfy; the frameworks are how we implement and audit it. Controls map to AI RMF Govern/Manage and the 800-53 AC/AU/IA families.

Differentiators

• Certified SDVOSB and VOSB (SBA VetCert) A certified service-disabled-veteran-owned small business focused specifically on federal AI governance engineering.
• U.S. patent application No. 64/080,982 (filed 2026) A deterministic cascade architecture placing a rules engine in authority over LLM output, with a dual-channel protocol carrying machine-readable state alongside natural-language generation.
• Sole architect of a production multi-provider LLM gateway At a major research university. The operational core of AI governance, built and run end to end.
• One audit, every language A language-invariant compliance architecture: rule-correctness is verified once, not re-certified per language.
• Principal 70% service-disabled U.S. Army veteran. CH-47 crew chief, Task Force Viking, Operation Iraqi Freedom (northern Iraq, 2003). CISSP (2022–2025), CompTIA Security+ (2019–2025).

Relevant Experience

• A production, multi-provider LLM governance gateway operated end to end at a major research university.
• Live deployments running the deterministic-enforcement harness, with per-exchange verdict logging and documented evaluation-scorer calibration.
• The company's first commercial title, an AI-driven narrative platform built on the patent-pending deterministic enforcement architecture, launching on Steam in late 2026.

Contact