Frequently Asked Questions

In fact, the formula is quite simple; to avoid encountering unexpected costs during the transition to the cloud, proper planning is essential. This will be possible through the experience and expertise of the chosen Cloud Service Provider. Simply having years of experience in the field is not sufficient. Additionally, a deep understanding of the customer's industry and the IT products they use is crucial. Pre-planning for Cloud Migration Strategies before the transition is a vital factor for success. Another critical factor is employing Expert Personnel. Periodic difficulties can arise, and encountering problems due to the inability to maintain expertise is a common issue. As a solution to all of these, if we are transitioning to the cloud with proper planning, migration strategy, and expert technical personnel, we should not expect any unexpected costs. As a leading local service provider, vMind guarantees that you will not encounter any negativity or surprises during the transition to the cloud, thanks to its expert team, years of experience, and various sector applications.

Data has become our most valuable asset today. Therefore, extra measures need to be taken for its security. The security measures taken are more important than where the data is stored. Borders are not significant for hackers; cybersecurity measures to prevent them are crucial. If you have established a more secure infrastructure than what a Cloud Provider with professional security teams provides 24/7, and if you can allocate a budget to ensure its continuity, then keeping it on local servers will ensure more safety. However, under today's conditions, it will be quite challenging both in terms of cost and finding expert personnel and ensuring their continuity. As a solution to the problem, it can be said that leaving the Security, which is one of the main tasks of a Cloud Service Provider, to the experts would be more appropriate. With vMind's servers equipped with high-security standards, global certifications, and 100% compliance with regulations, along with its expert team equipped with the latest information, vMind provides optimal solutions.

At vMind, data sovereignty and KVKK compliance are core principles. All customer data is hosted only in our Turkey‑based data centers that meet high security standards such as ISO 27001. We apply all technical and administrative measures required by regulation for processing, storing and destroying data, ensuring that your data remains protected, legally compliant and under your control.

Data ownership belongs to the customer. In case of contract termination or provider change, data export and migration processes are predefined. Open architectures and standard technologies minimize vendor lock‑in risk.

High availability is achieved through multi‑location data center architecture, redundant network infrastructure and defined disaster recovery scenarios. SLA agreements clearly define uptime targets and operational responsibilities.

DR services start with a business impact analysis. RPO and RTO targets are defined, followed by replication, backup and recovery scenarios. Contractual SLAs are established for critical systems.

First a technical and operational analysis is conducted. Rehost, refactor or hybrid alternatives are evaluated. Architecture is designed based on performance requirements and validated with a POC when necessary.

Resource usage is continuously monitored, capacity optimization is performed and cost reporting is provided. Pricing options in local currency help organizations maintain predictable budgets.

vMind provides end‑to‑end managed services beyond infrastructure. Monitoring, backup, security, database management and DevOps processes can all be managed by vMind if required.

vMind designs integrated architectures combining private and public cloud environments. Sensitive data can remain in sovereign infrastructure while global services can be integrated through hybrid models.

Network architectures are designed for centralized management, secure connectivity and low latency. Secure inter‑branch access and centralized monitoring ensure operational integrity.

Managed database services include installation, configuration, patch management, backup, performance monitoring and capacity planning.

Custom storage and network architectures are designed based on performance requirements. Hybrid models can be implemented to optimize critical workloads.

vMind offers not only technical infrastructure but also strategic planning, migration consulting and operational transformation services. Long‑term digital roadmap planning can also be supported.

Yes. Partnership and white‑label reseller models are available, enabling joint growth strategies through the channel ecosystem.

For regulated industries such as finance, healthcare and energy, specialized architectures are designed focusing on data sovereignty, auditability and SLA‑based reliability.

vMind positions itself as a technology partner that designs, operates and manages end‑to‑end digital infrastructures. With portvMind Public and Private Cloud platforms, managed services and a strong partner ecosystem, organizations gain secure, scalable and sustainable digital infrastructure.

Traditional IT support only kicks in after an incident has occurred, while managed services proactively detect and resolve issues before they arise through continuous monitoring. Additionally, it offers pre-defined response times guaranteed by SLA and a predictable cost structure.

Yes. All data is protected with security protocols, encryption standards, and access control policies defined under the SLA scope. Security vulnerabilities are proactively addressed through regular security audits and penetration tests. GDPR compliance is also managed within the service scope.

Absolutely. Thanks to the flexible package structure, small businesses can access enterprise-quality IT management with a service scope scaled to their needs. SMB-specific packages are designed to cover essential security and monitoring services without deploying the complex structures designed for large enterprises.

Yes. The monthly subscription model allows the service scope to be flexibly adjusted during growth periods or when workloads change. There are no pre-set exit periods or long-term lock-in.

Possibly, yes. The two models work in a complementary rather than conflicting manner. While your internal IT team focuses on strategic projects, innovation initiatives, and business development, routine operations, 24/7 monitoring, and incident management are handled by vMind.

No. The onboarding process is conducted with a seamless handover plan that allows existing systems to continue operating in parallel. Service quality is maintained during the average 2–4 week transition period.

SLA conditions are clearly defined in the contract. If committed response or resolution times are exceeded, service credit mechanisms are activated. All metrics are shared transparently via monthly reports.

AI infrastructure is all the layers needed to train and run models: GPU compute, data storage, network and orchestration. A GPU alone is not infrastructure. If the data layer is not solved, the cards sit idle waiting for data. More information...

No. GPU rental is only one part; it covers only compute power. AI infrastructure also includes storage, network, orchestration and monitoring layers. More information..

All workloads run in vMind data centers in Istanbul. Data is not sent abroad, and KVKK data residency rules are met. More information..

No. You can train your own model, run an open-source model, or fine-tune an existing model. The infrastructure works with any model. More information..

Yes. GPU resources are added to your current Private Cloud or Public Cloud environment. You do not need a new account or to move your data. More information..

The price depends on the card model, number of cards, usage time, commitment choice and the scope of managed services. Tell us what you need and we will prepare a custom offer. More information..

GPU infrastructure is the graphics processing units, the fast network that connects them, and the storage layer that feeds them data. In AI workloads, if these three layers are not sized together, the cards sit idle waiting for data. More information..

If your team has engineers who can run GPU infrastructure, choose GPU Rental. If not, choose GPU as a Service. If you have already picked a card model, you can continue from the model pages. More information..

Up to 8 cards on a single node and up to 64 cards in a multi-node cluster. For larger needs, we build a custom cluster. More information.. 

In vMind data centers in Istanbul with ISO 27001 certification. Data is not sent abroad. More information...

It is all the layers needed to train and run models: GPU compute, data storage, network and orchestration. A GPU alone is not infrastructure. More information..

It is technically possible, but it is not cost-effective. A training cluster has low usage when it runs inference, and the monthly cost can be up to 40% higher than expected. More information...

Yes. You can set up a GPU node pool on vMind Kubernetes Engine and run training and inference in the same cluster. More information..

The active dataset is kept on fast storage next to the GPU cluster. Checkpoints and archives are kept on Object Storage. All of it is in Istanbul. More information..

Yes. A model you trained or fine-tuned yourself can be hosted on vMind infrastructure. More information...

Fine-tuning trains the model again and is costly. RAG indexes your documents and lets the model read them. Use RAG to answer with company knowledge. Use fine-tuning to teach tone and format. More information...

Request and response content is not used to train models. Metadata logs for audit are kept for 30 days and can be turned off on request. More information..

We offer the Llama, Mistral and Qwen model families, plus models fine-tuned for Turkish. We also review requests for open-source models not on the list. More information...

No. The applications are ready. Setup is only about connecting your own sources. More information..

Standard setup takes about 6 weeks on average. The time depends on the size of the knowledge base and the number of source systems. More information..

Ready integrations are available for websites, intranet, email, Microsoft Teams and Slack. For company systems, we integrate through API. More information...

Answers come with a link to the source document. Wrong content is fixed at the source, and the index is updated. More information..

It means handing over the planning, compliance and operations of an AI project to a service provider. Your team builds the model; vMind runs the infrastructure and operations. More information..

Yes. Each of the three services can be bought separately. More information...

No. The result does not depend on any provider and can be used on other infrastructure too. More information...

99.99% monthly availability and a 15-minute response time for critical incidents. More information...

It depends on the card model, number of cards, usage time, commitment choice and the scope of managed services. Tell us what you need and we will prepare a custom offer. More information...

Yes. The minimum usage time is 1 hour. After that, you pay by the hour. More information...

Up to 8 cards on a single node and up to 64 cards in a multi-node cluster. For larger needs, we build a custom cluster. More information...

When you fully delete the resource, GPU billing stops. If you pause the resource, you pay only for the attached storage. More information...

In vMind data centers in Istanbul. Data is not sent abroad. More information...

Yes. You can set up a GPU node pool on vMind Kubernetes Engine. More information...

If your model fits in 80 GB of memory and the work is mostly computing, H100 is a good fit. If the model does not fit, or you work with long context windows, choose H200. More information...

No. You can use it without a commitment. With a yearly commitment, you get up to 20% off. More information...

Within 2 business days. For complex cluster requests, within 5 business days, together with an architecture proposal. More information...

The managed service layer is priced separately, based on scope and service level. Tell us what you need and we will prepare a side-by-side offer. More information...

Yes. Even though the vMind team manages it, your team gets console access with set permissions. More information...
 

There is no limit on models. You can run open-source models, models you built, or models you fine-tuned. More information...

Yes. We can take over the management of a GPU cluster in your own data center or at another provider. We do a technical review before the handover. More information...

99.99% monthly availability and a 15-minute response time for critical incidents. If we miss this, you get a service credit on your monthly fee. More information...

H100 has 80 GB of HBM3 memory, and H200 has 141 GB of HBM3e. H200 is better in memory size and bandwidth. H100 is more cost-effective for compute-heavy training. More information...

Up to 8 on a single node and up to 64 cards in a multi-node cluster. More information...

It depends on the number of cards, usage time and commitment choice. Fill in the contact form to get an offer. More information...

No. The cards are not split; they are fully dedicated to you. More information...

H200 has 141 GB HBM3e and 4.8 TB/s. H100 has 80 GB HBM3 and 3.35 TB/s. H200 is better in memory size and bandwidth. More information...

Up to 8 on a single node and up to 64 cards in a multi-node cluster. More information...

It can be used for both. Its advantage is clearer in inference workloads that need a lot of memory. More information...

It depends on the number of cards, usage time and commitment choice. Fill in the contact form to get an offer. More information...

Standard setups are ready in about 10 minutes on average. For custom clusters, we do a technical review. More information...

The active dataset is kept on fast storage next to the GPU cluster. Archive and checkpoint files are kept on Object Storage. More information...

Checkpoint files are saved regularly, so training continues from the last saved point. More information...

Yes. You can set up a GPU node pool on vMind Kubernetes Engine and manage training jobs there. More information...

It is the system where a trained AI model runs to answer live requests. More information...

It is technically possible, but not cost-effective. Large cards have low usage, and the monthly cost can be up to 40% higher than expected. More information...

p95 200 ms. We set this value together with you, based on model size and request complexity. More information...

The GPU pool scales automatically. Extra capacity does not need manual work. More information...

The new version is rolled out step by step. If there is a performance problem, we go back to the previous version. More information...

Yes. A model you trained or fine-tuned yourself can be hosted on vMind infrastructure. More information...

Request and response content is not used to train models. Metadata logs for audit are kept for 30 days and can be turned off on request. More information...

Kapasite ve kullanım hacmine göre belirlenir. Teklif için iletişim formunu doldurabilirsiniz.

Yes. Thanks to the standard API, your current integrations can be moved with only small changes. More information...

Fine-tuning trains the model again and is costly. RAG indexes your documents and lets the model read them. Use RAG to answer with company knowledge. Use fine-tuning to teach tone. More information...

PDF, Word, Excel and web content are supported. For other formats, we set up a conversion pipeline. More information...

Access is set for each document. Users get answers only from documents they are allowed to see. More information...

On vMind infrastructure in Istanbul with ISO 27001 certification. Documents are not sent abroad. More information...

It is for teams that build their own models, manage several model versions and move models to production. More information...

Yes. It is set up to work with popular open-source MLOps tools. More information...

The platform runs on Kubernetes-based infrastructure. It comes with vMind Kubernetes Engine. More information...

On vMind Object Storage in Istanbul, with versions and tags. More information...

Standard setup takes about 6 weeks on average. The time depends on the size of the knowledge base and the number of source systems. More information...

Ready integrations are available for websites, intranet, email, Microsoft Teams and Slack. For company systems, we integrate through API. More information...

Answers come with a link to the source document. Wrong content is fixed at the source, and the index is updated. More information... 

On vMind infrastructure in Istanbul with ISO 27001 certification. More information...

Yes. It works with handwritten forms. Accuracy depends on how clear the handwriting is. More information...

Data is sent through API to accounting, ERP and document management systems. More information...

Standard setup takes about 6 weeks on average. It depends on the number of document types and the scope of integration. More information...

On vMind infrastructure in Istanbul with ISO 27001 certification. Documents are not sent abroad. More information...

It can connect to company applications that have an API. We decide the integration scope together during setup. More information...

Critical steps are set up to need approval. Which steps need approval is decided during setup. More information...

Yes. Every step is logged and can be reviewed later. More information...

It depends on how complex the process is and the number of integrations. We share the timeline after a review. More information...

A standard assessment takes 4 weeks. A quick assessment focused on one use case takes 2 weeks. More information...

No. The result does not depend on any provider. You can use the architecture with another provider too. More information...

Finance, healthcare, manufacturing, retail, logistics and the public sector. We share references for your industry in a meeting. More information...

A list of use cases in order of priority, a target architecture, a cost plan and a step-by-step roadmap. More information...

When requests with personal data are sent abroad, data transfer rules apply. Running the model in Türkiye removes this risk. More information...

ISO 27001 certificate for information security management. We share the current list of our other certificates on request. More information...

Documents on technical and administrative measures, an access rights matrix, audit trail logs and a data processing inventory. More information...

Yes. We review your current setup, report compliance gaps and give you a plan to fix them. More information...

Yes. Even though vMind runs operations, your team gets access with set permissions. More information...

No. Model development stays with your team. vMind is responsible for keeping the model running. More information...

99.99% monthly availability and a 15-minute response time for critical incidents. If we miss this, you get a service credit. More information...

Yes. We do a technical review before the handover and prepare a migration plan. More information...