ZTABS deploys self-hosted AI infrastructure for Houston enterprises across energy operations, aerospace engineering and defense contractors. Self-hosted deployment runs open-weight language models on customer-controlled hardware. Whether inference and data remain within the organisation's security boundary depends on the actual configured full-stack data paths including inference endpoints, embedding services, vector storage, telemetry, update channels and admin tools; on-premises hardware alone does not inherently guarantee zero external transmission. The deployment preserves operational control for environments where cloud AI API access is restricted by contract, regulation or organisational policy.
Houston self-hosted AI demand spans energy operations (engineering documentation, operational knowledge management), aerospace engineering (design assistance, documentation analysis) and defense contractors (CUI-handling tools, engineering support within classified or controlled environments). ZTABS is remote-first by design, with an office in Houston, Texas. Our self-hosted AI deployment services cover infrastructure deployment across markets.
Deployment scenarios and requirements
| Need | What we build | Key requirement |
|---|
| Energy operations | Engineering documentation AI, operational knowledge tools | Approved engineering authority, safety interlocks |
| Aerospace engineering | Design analysis AI, documentation search, code assistance | ITAR encrypted storage, export control assessment |
| Defense contractors | CUI-handling AI tools, engineering support | NIST SP 800-171 Rev 2, DFARS 252.204-7012 |
| Healthcare (Texas Medical Center) | Clinical documentation AI, research tools | HIPAA BAA, PHI safeguards |
What self-hosted means
Self-hosted deployment runs open-weight language models (such as Llama, Mistral, Qwen or DeepSeek families) on hardware the organisation controls. The organisation retains control over model weights, inference data and operational configuration. Self-hosting addresses scenarios where cloud AI APIs are contractually prohibited, where data classification requires controlled infrastructure, or where the organisation's unit economics favour on-premises inference at their query volume.
What self-hosting does not automatically provide. On-premises hardware, open weights and local GPU do not by themselves mean no data leaves the organisation. The full chain must be mapped: inference endpoints, embedding services, vector storage, logs, backups, telemetry, model update channels, admin support tools, external tool connections, identity providers and network egress. Each component's actual data flow is documented in the deployment architecture.
Model licences. Open-weight models have specific licence terms and acceptable use policies that differ from unrestricted open-source licences. Current model selection considers licence terms, runtime features and use conditions applicable to the engagement's intended purpose. Licence terms may restrict certain uses or require attribution.
Planning self-hosted AI for Houston energy, aerospace or defense?
Tell us about your infrastructure, security requirements and use cases. We start with environment assessment and workload analysis.
CUI safeguarding and NIST SP 800-171
Defense contractors handling Controlled Unclassified Information (CUI) must safeguard it under DFARS 252.204-7012 and NIST SP 800-171 Rev 2. CUI is unclassified information that requires safeguarding; it is distinct from classified information (Secret, Top Secret) and SCI/SAP programmes which have their own handling requirements under separate authorities.
CMMC current status. The DoD suspended CMMC Phase II on 13 July 2026, pausing the contractual transition to third-party assessment requirements including the November 2026 transition timeline. Phase I self-assessments remain required. During the suspension, solicitations may require only CMMC Level 1 (Self) or Level 2 (Self) assessments; self-assessments and select government-led assessments continue as applicable. DFARS 252.204-7012 and NIST SP 800-171 Rev 2 obligations remain in effect. A CMMC Reform Task Force has been established for a 60-day review.
The self-hosted deployment supports the contractor's safeguarding obligations by implementing the applicable NIST SP 800-171 controls for the AI infrastructure. Security readiness support is distinct from certification or authorisation; the contractor obtains any required assessment or authorisation through the applicable process. ZTABS does not offer security clearances, SCIF access, ATO issuance or assessment authority. Actual named personnel eligibility, approved environment and customer-authorising official requirements are defined by the contractor's programme.
ITAR encrypted storage conditions
22 CFR 120.54(a)(5) provides that sending, taking or storing unclassified ITAR technical data is not an export when: the data is secured using end-to-end encryption with cryptographic modules compliant with FIPS 140-2 or successors and current NIST-aligned key management and procedures (or other cryptographic means providing at least AES-128 comparable security strength), the means of decryption are not provided to any third party, the data is not intentionally sent to a person in or stored in a section 126.1 proscribed country, and the data is not sent from a section 126.1 proscribed country. Data in transit via the internet is not deemed stored in a country it transits.
This exception covers unclassified technical data only; classified ITAR data is not covered regardless of encryption. End-to-end encryption is defined as cryptographic protection between the originator (or originator's in-country security boundary) and the intended recipient (or recipient's in-country security boundary); the intended recipient must be the originator, a US person in the United States or a person otherwise authorised to receive the technical data. The ability to access technical data in encrypted form satisfying these conditions does not by itself constitute release or export. The exception does not authorise otherwise prohibited release or decrypted inference access by unauthorised persons; use, end-user and export assessments remain separate from the encrypted storage and transport condition. Providing access information enabling a foreign person to access unencrypted technical data is a controlled release under 22 CFR 120.56 and requires authorisation.
Self-hosted AI inference on ITAR-controlled data requires that the inference environment satisfies the applicable encryption, access control and destination conditions. The deployment architecture documents the actual data flow for ITAR-controlled information across all system components.
Energy operations deployment
Houston's energy sector creates demand for self-hosted AI supporting engineering documentation, operational knowledge management and data analysis.
Engineering documentation AI. Deploys language models with RAG capabilities for engineering document search, specification analysis and lessons learned retrieval. Energy production, drilling and operational decisions require approved engineering authority with independent safety interlocks. The AI provides advisory analysis; it does not control physical processes or replace qualified engineering assessment.
PLM and DOORS integration. Self-hosted AI can integrate with product lifecycle management systems and requirements management tools where API access and data entitlements are confirmed. Write operations to engineering systems require explicit approved authority with change control.
Aerospace and defense deployment
Houston's aerospace sector (including NASA JSC contractors and defense aerospace) creates demand for self-hosted AI within controlled environments.
Deployment tiers. Self-hosted deployments may span multiple security tiers depending on the organisation's environment: unclassified corporate networks, CUI-handling environments with NIST SP 800-171 controls, and higher classification environments where applicable. Each tier has its own access controls, personnel requirements and data handling procedures. The deployment architecture addresses each tier's specific requirements.
SSP documentation. System Security Plans document the security controls implemented for the AI infrastructure. The SSP scopes the system boundary, identifies the applicable security requirements and documents the control implementation. SSP development and maintenance support is part of the deployment engagement where agreed.
Medical Center deployment
Houston's Texas Medical Center creates demand for self-hosted AI supporting clinical documentation and research.
HIPAA responsibilities. Self-hosted deployment does not eliminate HIPAA Business Associate requirements. Per HHS cloud computing guidance, a provider that creates, receives, maintains or transmits PHI on behalf of a covered entity is a Business Associate under all three covered entity categories. BAA terms, Security Rule safeguards and applicable Privacy Rule provisions apply. OCR does not certify products for HIPAA compliance. The specific BA role and PHI scope are defined per engagement.
Supply chain and update management
Approved supply chain. Model weights, runtime dependencies and system components follow approved procurement with provenance review. Pinned versions, cryptographic signatures and dependency review support supply chain integrity. Staging environments with evaluation against held-out test sets precede production deployment. Rollback capability is maintained.
Ollama and runtime tools. Tools like Ollama provide convenient local inference runtime but are not single-binary zero-dependency packages with inherent zero data exfiltration. Per the Ollama FAQ, cloud-connected features and web search capabilities exist. The OLLAMA_NO_CLOUD environment variable disables cloud-connected features but is not a universal telemetry or network egress kill switch; runtime network controls, firewall rules and update channel configuration are separately assessed and tested for the deployment's security requirements.
Illustrative Example
Hypothetical: Houston defense contractor self-hosted AI deployment
A Houston defense contractor handling CUI across engineering and operations departments needs self-hosted AI to support document analysis, code assistance and engineering knowledge retrieval within the organisation's CUI-handling environment.
The deployment would install open-weight language models on the contractor's GPU infrastructure within the CUI network boundary. RAG capabilities would index approved engineering documentation, specifications and procedures. Engineers and analysts would query the AI through an internal interface respecting the existing access control model.
Assumptions: the contractor's CUI environment has existing NIST SP 800-171 controls in place, GPU hardware procurement follows the contractor's approved acquisition process, model weights and dependencies are transferred through the facility's approved media transfer procedures, DFARS 252.204-7012 obligations are addressed within the contractor's existing SSP, and actual personnel eligibility and environment approval requirements are defined by the contractor's programme office. Residual risks include model update lag in restricted environments, inference quality variation across engineering domains, GPU capacity planning for concurrent users and operational knowledge transfer timelines. No guarantee of specific productivity improvement, code development speed increase or financial return.
This is a hypothetical scenario for illustration. Actual scope, timeline and cost depend on your requirements.
Cost drivers for self-hosted AI
Self-hosted AI deployment has no published cost guide on this site. Cost is driven by factors that vary significantly across engagements. Rather than quoting numeric tiers, the following framework helps buyers assess scope.
Model and workload factors. Model parameter count, context window requirements, KV cache sizing, precision and quantisation choices, batching strategy and concurrency targets determine the GPU infrastructure requirements. Workload benchmarking with representative queries establishes actual throughput and latency characteristics rather than assumed specifications.
Infrastructure factors. GPU procurement or lease decisions, high-availability and redundancy requirements, power and cooling capacity, network configuration, storage for model weights and vector indexes, and backup infrastructure. For restricted environments, approved procurement channels and supply chain review add lead time.
Security and compliance factors. Security readiness engineering (NIST SP 800-171 controls, ITAR encrypted storage implementation, HIPAA safeguards), independent assessment and authorisation processes (scoped and priced separately), SSP documentation and ongoing compliance maintenance. Government authorisation processes are separate from engineering scope.
Integration factors. Connection to existing enterprise systems (PLM, DOORS, ERP, document management), API development, authentication integration and data migration. Each integration's actual API access, licences and data entitlements must be confirmed.
Operations and handover factors. Operations documentation, monitoring setup, alerting configuration, model update procedures, security patch management, capacity planning and knowledge transfer. Actual operations resources, service window and staffing depend on deployment scale and security requirements.
Procurement milestones. Hardware procurement, environment preparation, model selection and benchmarking, security readiness assessment, deployment and integration, evaluation and acceptance testing, operations handover. Each milestone has measurable acceptance criteria agreed at engagement start.
Need to scope self-hosted AI for your Houston facility?
Share your environment, security requirements and use cases. We provide a deployment assessment covering infrastructure, compliance and operational scope.
Proposal evaluation criteria
- Relevant self-hosted AI deployment experience across controlled environments including CUI, ITAR and HIPAA with supporting project evidence
- Infrastructure design including GPU workload benchmarking, high-availability architecture and capacity planning with actual performance evaluation
- Security readiness engineering with clear NIST SP 800-171 control implementation, ITAR encrypted storage conditions and supply chain provenance, with independent assessment responsibilities separately identified
- Model evaluation methodology with held-out test cases, domain-specific benchmarking and actual selected model task evaluation
- Transparent scope, milestones, change control, operating fees and ongoing support terms
- Provider team identifies who delivers deployment engineering, trains agreed users on AI tools and operations procedures, delivers technical and operations documentation, and provides ongoing support; customer team identifies who provides environment access, validates domain-specific model outputs, accepts delivery milestones, approves security controls and handles operational responsibility post-handover; both parties agree staff and user languages, documentation scope, translation responsibilities and handover materials
How delivery works
ZTABS is remote-first by design, with an office in Houston, Texas. Central time zone coordination for Houston-based clients. For environments requiring on-site presence, on-site access is separately agreed based on facility requirements and personnel eligibility.
Environment assessment and workload analysis. The engagement begins with infrastructure review covering existing GPU resources, network architecture, security boundary, data classification levels and integration targets (PLM, DOORS, ERP, document management). Workload benchmarking with representative queries establishes throughput and latency baselines.
Model selection and licence review. Candidate open-weight models are evaluated against domain-specific held-out test sets covering energy engineering, aerospace and defense terminology. Model licence terms, acceptable use policies and runtime requirements are reviewed for compatibility with the engagement's intended purpose and environment restrictions.
Security readiness and permission-boundary testing. NIST SP 800-171 controls, ITAR encrypted storage conditions and HIPAA safeguards are implemented as applicable. Test sets verify access controls, CUI boundary enforcement and data isolation. Negative tests confirm the AI refuses out-of-scope queries and does not surface restricted data across classification or project boundaries.
Pilot and acceptance. Pilot deployment targets a single agreed use case with acceptance testing including task accuracy, retrieval correctness and security control verification. Rollout extends to additional use cases with rollback capability at each stage.
Training and handover. Agreed user training sessions, domain validation methodology transfer to customer operations team, and complete operations documentation including SSP updates, monitoring configuration, model update procedures and support escalation contacts.
When ZTABS is not the right fit
- Turnkey GPU hardware procurement for this engagement. We design infrastructure requirements; hardware procurement follows the organisation's approved channels.
- Security clearance or SCIF access for this engagement. Personnel eligibility and environment access are the contractor's programme responsibility.
- ATO issuance or C3PAO assessment for this engagement. Authorisation and assessment are obtained through the applicable process; ZTABS supports security readiness engineering.
- Ongoing managed hosting or admin for this engagement. Identify internal operations team for post-handover responsibility; ongoing retainers are separately agreed.
- On-site embedded team for this engagement. ZTABS is remote-first by design. On-site presence is separately agreed where required.
White-label AI products
ZTABS builds custom white-label AI products for clients to brand and resell. Houston enterprises can commission white-label self-hosted AI platforms for distribution.
Data governance
Each engagement agrees the data classification scope, processing boundaries and retention requirements for the self-hosted deployment. Model weight provenance, inference data retention, log management and deletion procedures are documented. Deletion and access revocation across source documents, copied chunks, index entries, embeddings, cached results, log entries, backup copies and any training or fine-tuning artifacts requires coordinated procedures with verified actions at each layer. Lawful retention obligations and legal holds are distinguished from active retrieval; access is restricted rather than data deleted where retention duties exist, with propagation lag disclosed and access restricted during lag. Trained model weights may retain residual memory that is not automatically reversible through deletion; deleting source records does not erase model-weight memory. The engagement specifies deletion scope, verification method and residual risk acceptance. Contractual responsibilities and actual data paths define the deletion scope.
Monitoring and operations
Infrastructure monitoring. GPU utilisation, memory pressure, inference latency, throughput, storage utilisation and network health.
Security monitoring. Access logs, authentication events, data transfer logs and anomaly detection within the deployment's security boundary.
Model quality. Evaluated against held-out test sets with agreed acceptance criteria. Drift detection and periodic re-evaluation against updated benchmarks.
Handover
Deployment handover includes infrastructure runbooks, model configuration, security documentation (SSP where applicable), monitoring dashboards, update procedures and escalation contacts.
Ready to deploy self-hosted AI in Houston?
Share your infrastructure, security classification and use cases. We start with environment assessment and workload analysis.