Terms of Service DEVANUM

1. Acceptance of Terms

Welcome to DEVANUM. By engaging DEVANUM for specialized AI Engineering, Large Language Model development, agentic workflows, scalable cloud infrastructure, or intelligent software systems, you (“Client”) agree to be bound by these Terms of Service (“Terms”).

Please read these Terms carefully. If you do not agree to these Terms, you may not access our services or engage DEVANUM for engineering projects.

These Terms, together with any signed Statement of Work (SOW), Master Service Agreement (MSA), or project proposal, constitute the entire legal agreement between the Client and DEVANUM.

2. Scope of Services

DEVANUM provides highly technical AI Engineering services, which are strictly defined in the signed Statement of Work (SOW) or MSA for each engagement. Services may include:

• Custom LLM development, fine-tuning, and architectural optimization.
• Design and implementation of RAG (Retrieval-Augmented Generation) platforms.
• Construction of multi-agent autonomous AI systems and workflows.
• Provisioning of specialized GPU infrastructure and MLOps pipelines.

Any modifications to the Scope of Services must be approved in writing by both parties via a formal Change Order.

3. Payment, Compute Costs & Compute Credits

Payment terms are outlined in the specific MSA or project SOW.

Compute Infrastructure Costs

Strict Responsibility: Client acknowledges that AI engineering requires significant specialized infrastructure (e.g., A100/H100 GPU clusters). The Client is strictly responsible for all direct cloud infrastructure, compute costs, and API usage fees incurred during the development, training, fine-tuning, and deployment phases of the project, unless expressly noted otherwise in the SOW.

DEVANUM Compute Credits

DEVANUM may, at its discretion, provide limited **Compute Credits** for initial experimentation or prototyping. These credits:

• Are non-transferable, have no cash value, and expire as detailed in the proposal.
• Can only be used for DEVANUM managed infrastructure.
• Are subject to fair-use policies and availability.

4. Intellectual Property Rights

Our approach to IP is structured to ensure Client ownership of final delivery while protecting DEVANUM’s specialized internal tooling.

Client Ownership

Upon full payment of all invoices related to the project, the Client will own all developed software code, custom training datasets, and the finalized optimized model weights and parameters ("Deliverables"), specifically created for the Client as part of the Services.

DEVANUM Internal Tooling & Accelerators

DEVANUM retains all ownership, rights, and interest in and to its pre-existing internal tools, software accelerators, training frameworks, proprietary algorithms, and knowledge repositories used to develop the Deliverables ("DEVANUM Background IP").

5. Client Data Usage

DEVANUM requires access to Client Data to perform the services, particularly for training, evaluating, and fine-tuning AI models.

Model Training Limitation

DEVANUM guarantees that Client Data will **NEVER** be used to train DEVANUM’s proprietary models, internal accelerators, or another client's custom AI systems. Client Data is used strictly for the project identified in the SOW.

6. Confidentiality

Both parties agree to protect the Confidential Information of the other. For DEVANUM, this includes proprietary technical frameworks, model optimization techniques, and pricing strategies. For the Client, this includes proprietary business data and trade secrets.

These obligations remain in force for five (5) years following the termination of services.

7. AI Warranties & Nondeterminism

Nondeterministic Nature of AI: Client acknowledges that Large Language Models, Generative AI systems, and Autonomous Agents are nondeterministic. Outputs can vary and cannot be guaranteed to be 100% accurate, safe, or reliable for any specific purpose. Models may produce incorrect or unexpected results ("hallucinations").

NO WARRANTY: DEVANUM DELIVERS ALL AI SERVICES, MODELS, SYSTEMS, AND AGENT WORKFLOWS "AS IS" AND "AS AVAILABLE." DEVANUM EXPRESSLY DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, AND WARRANTIES THAT AI OUTPUTS WILL BE ERROR-FREE OR HALLUCINATION-FREE.

8. Limitation of Liability

IN NO EVENT SHALL DEVANUM’S AGGREGATE LIABILITY ARISING OUT OF OR RELATED TO THESE TERMS, WHETHER IN CONTRACT, TORT, OR UNDER ANY OTHER THEORY OF LIABILITY, EXCEED THE TOTAL AMOUNT PAID BY CLIENT TO DEVANUM IN THE SIX (6) MONTHS PRECEDING THE INCIDENT GIVING RISE TO THE LIABILITY.

DEVANUM SHALL NOT BE LIABLE FOR ANY INDIRECT, SPECIAL, INCIDENTAL, PUNITIVE, OR CONSEQUENTIAL DAMAGES, INCLUDING LOSS OF PROFITS, DATA, COMPUTE RESOURCE COSTS, OR BUSINESS INTERRUPTION.

9. Term & Termination

The term of this Agreement is defined in the MSA or project SOW. Either party may terminate for convenience with sixty (60) days’ written notice.

DEVANUM may immediately suspend or terminate services if the Client fails to pay undisputed invoices for compute resource costs within ten (10) days of the due date.

10. Governing Law

These Terms and any non-contractual obligations arising from them shall be governed by and construed in accordance with the laws of the **State of Delaware, USA**.

Both parties agree that any dispute arising under this Agreement shall be resolved through binding arbitration in Wilmington, Delaware, under the rules of the American Arbitration Association (AAA).