From learning the technology to testing a business decision.

A manufacturing team puts its first digital twin to the test.

• Five-day initial training
• Follow-up office hours
• Proof of concept delivered in six months

A project ready to start. An environment that was not.

An advanced manufacturing team had identified a practical first use for digital twins: represent a manufacturing line virtually and explore how the technology could support its work. The obstacle was getting started inside an enterprise with demanding security requirements.

Bringing unfamiliar software and external expertise into the corporate environment required a lengthy assessment process. The team needed a way to learn and experiment while those constraints were addressed.

Training organized around the proof of concept

UME.Studio worked with the team to understand the project's goals, existing skills, and security constraints. Those requirements shaped a focused curriculum rather than a broad introduction to every feature of the technology.

The team began in an isolated virtual training environment, accessed through its browsers and separated from corporate networks. Five days of initial instruction were followed by office hours and practical support as the engineers applied NVIDIA Omniverse to their project.

Evidence the team could build on

The team completed its proof of concept in six months. It gave internal stakeholders a concrete demonstration of digital twin potential and showed that the engineers could develop the skills needed to continue.

At the time the engagement was documented, the team was working toward a more advanced second proof of concept, with a production pilot as the intended next stage.

Technology and delivery

NVIDIA Omniverse; UME.Studio virtual training workstations; custom curriculum; live instruction and follow-up office hours.

Turning digital twin interest into working demonstrations.

• 18 practitioners
• Six-week program
• 90% completed and presented a final capstone

Client interest was growing faster than practical experience.

A global professional services firm had followed digital twin technology for years. As client interest grew, its internal technology community needed a way to turn that familiarity into practical capability.

The participants brought different backgrounds and levels of technical experience. Analysts, consultants, and managers needed to learn together while continuing their client work across a distributed team.

A working project became the learning path.

UME.Studio designed a six-week live practicum around a final capstone. Participants applied visualization, simulation, and data-connection concepts in NVIDIA Omniverse, using prepared virtual workstations and instructor support.

Breakout sessions and individual guidance gave participants room to progress from different starting points. The program ended with teams running their applications live for management, making the learning visible through working demonstrations.

One team developed a healthcare simulation in which vials passed through a virtual scanner and were analyzed and sorted using an NVIDIA NIM integration. It was a capstone demonstration of connected technologies, rather than a claim of a validated pharmaceutical production system.

Capability that could be seen and shared

Eighteen practitioners joined the program. The source case study reports that 90% successfully built and presented a final capstone. Alongside those projects, the engagement created an internal community with a shared practical foundation for further digital twin work.

Technology and delivery

NVIDIA Omniverse; NVIDIA NIM in one healthcare capstone; UME.Studio virtual workstations and collaborative sessions; live instruction and project presentations.

Before changing the fulfillment line, learn to work with the whole facility.

• Two company engagements
• Mechanical engineering teams trained
• Preparation for a first proof of concept

A new station changes more than one section of conveyor.

Two fulfillment and distribution companies wanted to explore digital twins of their existing facilities. Their engineers were considering how revised layouts, inspection stages, and automation could fit into established fulfillment lines.

That meant understanding the facility as a connected environment. A proposed station needed to be considered alongside neighboring equipment, material flow, and access for the people working around it.

The first challenge was making a facility-sized dataset usable.

UME.Studio trained mechanical engineering teams in NVIDIA Omniverse and OpenUSD, with particular attention to managing the large datasets involved in representing an entire facility.

The work connected the technology to the engineers' own questions: how to organize a facility representation, explore equipment placement, and understand where inspection and automation might contribute value. OpenUSD provided the scene-management foundation for working with that scale of information.

A starting point for a useful proof of concept

These engagements were the teams' entry point into the technology. They built an understanding of what was possible and began the work toward proofs of concept relevant to their existing facilities.

The outcome at this stage was a foundation for further engineering work. Completed layout optimization, deployed automation, and measured throughput improvements are not claimed.

Technology and delivery

NVIDIA Omniverse; OpenUSD; tailored training for mechanical engineers working with large facility datasets.

Train vial inspection on the defects that are hard to collect.

• Customer-specific synthetic data for inspection-model training and refinement

The rare defect is often the one you most need to recognize.

Vision inspection on a high-speed pharmaceutical line depends on relevant examples of acceptable and defective products. Collecting those examples from production can be disruptive, particularly when defects are uncommon.

In work with a pharmaceutical customer, UME.Studio used synthetic data generated to match the customer's own equipment to support training and refinement of vision models.

Make the training data relevant to the inspection task.

A useful synthetic image must represent the conditions the model will encounter. Equipment geometry, viewpoint, illumination, and product appearance all influence what a defect looks like to a camera.

The accompanying visual explores a crack near a vial's glass rim beneath its cap. It makes the inspection problem tangible: detect a small defect, connect that detection to a specific vial, and separate it from the main flow. This is an editorial illustration of the application, rather than a claim about the exact customer implementation.

The decision is readiness for the line.

The purpose of this work is to help determine whether an inspection model is accurate enough and fast enough for its intended operating conditions. The supplied account does not establish a measured accuracy improvement, rejection rate, or completed production deployment.

Confirmed technology

Customer-specific synthetic data generation and vision-model training. Exact customer software and model names are withheld or unconfirmed.