From University Friendship to Pioneering AI and Humanoid Robotics
How Cosmin Covaciu and Bogdan Pintea built Alba Vision — a Romanian-born startup advancing AI and humanoid robotics for the future.
Chapter I – A Vision That Endured Through Time
In an era where technology is reshaping human boundaries and artificial intelligence is redefining entire industries and societies, the story of two engineering graduates from Romania becomes a compelling testament to what human talent can achieve when it is driven by vision, resilience, and a shared sense of purpose. Cosmin Covaciu and Bogdan Pintea are not merely former classmates or engineers with international expertise. They are the co-founders of a rising venture that aims to place Romania on the European innovation map for humanoid robotics and applied AI: Alba Vision. Alba Vision is more than a startup. It is the embodiment of a mission: to accelerate the responsible integration of AI and humanoid robots into critical sectors like retail, smart manufacturing, education, urban services, and healthcare. The project is founded on three core values — sustainability, accessibility, and modular innovation — with an ambition to contribute to a new paradigm of human-machine collaboration. The genesis of Alba Vision lies in something profoundly human: a friendship forged during university studies, tested by time, and later transformed into a professional alliance rooted in shared principles and complementary skills. Cosmin and Bogdan both graduated from the Technical University of Cluj-Napoca, Romania, one of the country’s leading academic institutions for engineering and computer science. Their early collaboration on robotics competitions, AI modeling challenges, and applied electronics set the foundation for a lasting intellectual partnership. Fast-forward over a decade, and the two have reconvened with a new mission: to build a scalable AI-first company that uses advanced robotics platforms and NVIDIA technology to tackle real-world automation challenges. In this first chapter, we explore the origins of their journey, their academic and professional growth, and the founding values that shape the Alba Vision initiative today — an initiative actively preparing to join the NVIDIA Inception program and to become a key contributor to the next generation of AI and robotics solutions in Central and Eastern Europe.
Chapter II – Where It All Began: Academia and Friendship
In the early 2000s, Romania was beginning to emerge as a vibrant ecosystem for STEM education. The Technical University of Cluj-Napoca had already built a reputation for excellence in electrical engineering, automation, and computer science. It was within this environment that Cosmin Covaciu and Bogdan Pintea began their academic journey. Both were drawn to systems engineering, robotics, and software development — fields that, at the time, were only beginning to converge in practical applications. Cosmin showed early interest in embedded systems and intelligent control algorithms, while Bogdan leaned toward applied mechanics and industrial systems integration. Their academic paths frequently intersected during laboratory courses, innovation workshops, and national student competitions. What began as a collegial acquaintance soon evolved into a collaborative partnership. Together, they participated in: Autonomous robot contests hosted by the Romanian Robotics Society; Research projects focused on adaptive control and mechatronics; Hackathons involving real-time simulation, object detection, and decision logic; Early-stage AI frameworks that preceded modern deep learning libraries. The dynamic between them was natural: Cosmin brought a systems-oriented mindset and a passion for software innovation, while Bogdan added a deep mechanical understanding and an eye for industrial applications. Their complementarity would later prove essential when launching Alba Vision.
Chapter III – International Experience, Local Commitment
After graduation, both took different professional paths, yet their interests remained parallel. Cosmin Covaciu built a multidisciplinary career that spanned industrial automation, IoT systems, and digital transformation. As CEO of Red Internet Sales SRL and later coordinator of technology projects in the energy and healthcare sectors, he gained firsthand experience in integrating cloud platforms, AI models, and sensor networks into mission-critical operations. A strong advocate of open-source technologies and edge computing, Cosmin also developed a deep appreciation for NVIDIA’s ecosystem — particularly Jetson modules, TensorRT acceleration, and CUDA-based optimization — which now inform the technical direction of Alba Vision. Bogdan Pintea, on the other hand, focused his career on mechanical design and production systems, contributing to several advanced manufacturing projects across Europe. His expertise in CAD-CAM, industrial robotics, and rapid prototyping became invaluable in the age of hardware-software co-design. Bogdan’s background includes working with six-axis robotic arms, collaborative robots (cobots), and AGV (automated guided vehicle) platforms, aligning perfectly with the needs of Alba Vision's robotics product development. Despite working in different sectors and cities, the two remained connected. Through discussions on emerging trends in AI, robotics, and sustainability, it became increasingly clear that they shared a vision for the future — one in which humanoid robots are no longer science fiction, but integrated co-workers in both physical and digital environments.
Chapter IV – Founding Alba Vision: A Startup with a Mission
In 2024, the time was right. Both founders had accumulated over a decade of experience in their respective fields, and the technological maturity of tools like ROS 2, Isaac Sim, Jetson Orin, and Transformer-based AI models made their idea viable. Thus, Alba Vision was born — an independent, founder-led technology startup headquartered in central Romania, with a core team of experts in: Robotics software engineering; Embedded AI systems; Human-machine interaction; Real-time simulation and reinforcement learning. The mission was ambitious: to design, simulate, and test modular humanoid robots that are: AI-native: Designed from the ground up to leverage perception, prediction, and learning; Modular and accessible: Components that can be adapted to different industries or user needs; Sustainable by design: Using recyclable materials and low-energy components; Edge-ready: Operable using edge AI and local compute, minimizing latency and increasing privacy. Their roadmap includes the development of three prototype families: Retail & logistics humanoid assistant – Capable of shelf restocking, aisle navigation, and customer interaction; Educational humanoid robot – An affordable, safe platform for STEM learning and AI curriculum integration; Care & support robot – A system tailored for eldercare and rehabilitation environments. This strategic focus positions Alba Vision not only as a robotics startup but as a cross-domain platform for scalable AI deployment in the real world.
Chapter V – Aligning with NVIDIA Inception
As Alba Vision formalized its development roadmap, it became evident that success would depend on its ability to integrate best-in-class hardware and software solutions for robotics and AI. That is where NVIDIA technology plays a key role. The startup has already begun testing and validating several of NVIDIA’s offerings, including: Jetson Orin NX and Nano for low-latency edge AI inference in real-world robotic applications; Isaac Sim and Isaac Lab for high-fidelity reinforcement learning training in simulated 3D environments; Omniverse for collaborative design, testing, and visualization of robot behaviors; CUDA + cuDNN pipelines for custom neural network optimization; TensorRT to accelerate real-time decision models onboard robots. By leveraging these tools, Alba Vision ensures that its prototypes are not only visionary but also technically competitive, scalable, and aligned with industry standards. The startup’s goal is to transition from isolated tests to multi-agent deployments in retail chains, logistics hubs, and smart city pilot zones within 24 months. More importantly, Alba Vision aligns philosophically with NVIDIA’s emphasis on: Accelerated computing for real-world impact; AI democratization; Edge-to-cloud robotics architectures; Interoperability across platforms. As such, applying for NVIDIA Inception is not a symbolic gesture, but a strategic milestone in Alba Vision’s evolution. It signals the company's readiness to enter the global conversation on robotics innovation and to become a serious contributor to the NVIDIA startup ecosystem.
Chapter VI – From Prototype to Platform
With a clear mission and strong technological foundation, Alba Vision began work on its first robotic prototype series. Instead of pursuing a single-purpose robot, the team envisioned a modular humanoid framework — a scalable architecture capable of adapting to various industry needs through hardware and software extensions. Key architectural principles include: Modularity: Each limb, joint, and sensor can be independently replaced or upgraded; AI-by-design: Every subsystem is built around data — for perception, prediction, planning, and control; Cloud-native orchestration: Robots can operate autonomously or as part of a fleet managed remotely; Edge inference with Jetson: Low-latency response using NVIDIA Jetson Orin modules for real-time decision-making. In practice, this means Alba Vision's robots can shift roles dynamically: A humanoid assistant trained in Isaac Sim for pick-and-place tasks can be quickly reconfigured to operate in education or retail; The same perception model used for object recognition can be reused across warehouse and healthcare environments; New modules (e.g., voice assistant, touchscreen, gesture sensor, barcode scanner) can be added on demand. This flexibility is not theoretical. The founders have already developed multiple simulations and hardware testbeds using ROS 2, MoveIt, Isaac Sim, and NVIDIA hardware, integrating: 6-DoF robotic arms with slip-ring-enabled infinite rotation; Stereo depth cameras for 3D scene reconstruction; Voice/gesture interfaces for human interaction; Battery management systems and actuated wheels for mobile autonomy. The result is a robotics platform that can serve as a learning assistant, warehouse worker, or shop-floor support unit — all from the same codebase and mechanical core.
Chapter VII – Building the Dream Team
To deliver on this ambitious roadmap, Alba Vision began building a multidisciplinary team of engineers, developers, researchers, and business experts. At the core are: AI & Software Engineers: Specializing in PyTorch, TensorFlow, Isaac SDK, and CUDA. Robotics Designers: Experts in CAD, kinematics, and real-world mechanics. Embedded Systems Specialists: Developing firmware and communication protocols. Cloud & Edge Architects: Managing deployment pipelines and fleet orchestration. Ethics and Accessibility Consultants: Ensuring inclusive design and responsible AI. In the short term, the team is focused on two primary hires: Reinforcement Learning Researcher with Isaac Lab experience; Mechanical Prototyping Engineer to iterate robot chassis for high-load operations. The long-term strategy includes: Collaboration with academic institutions (including former mentors from Cluj-Napoca); Hosting student internships in AI & Robotics; Participation in EU-funded research programs and regional innovation networks. As a startup aiming to join NVIDIA Inception, Alba Vision not only seeks access to technology but also to community and mentorship — to validate its research and expose its team to world-class development practices.
Chapter VIII – Partnerships and Deployment Strategy
A robotics company cannot exist in a vacuum. That is why Alba Vision places a strong emphasis on partnership development. Currently, the team is engaged in preliminary discussions with: Retail operators interested in humanoid robots for inventory and customer interaction; Smart city platforms aiming to integrate humanoid assistants in public service delivery; Universities and vocational schools looking for safe, modular robots for education and experimentation; Healthcare providers exploring low-cost robotic caregivers for eldercare and rehabilitation. In addition, the team is actively seeking technical partnerships with: NVIDIA for access to Inception resources; Cloud providers (Azure, AWS) for hybrid orchestration of simulations and robot control; Industrial hardware suppliers for high-torque actuators, precision encoders, and AI-enabled cameras; EU robotics alliances for certification, funding, and policy compliance. One of Alba Vision’s key objectives for 2025 is to pilot its humanoid platform in three domains: Smart Retail (human-robot collaboration): Shelf auditing, client support, and stock relocation; STEM Education (robotics-as-a-service): Full-body robots with programmable AI modules for schools; Rehabilitation Centers: Guided movement exercises with real-time patient feedback. Each of these scenarios is supported by NVIDIA-powered perception, planning, and simulation components.
Chapter IX – Responsible Innovation
Beyond engineering and performance, Alba Vision is guided by a commitment to social responsibility. The startup sees humanoid robotics not as a luxury or curiosity but as a transformational tool that can: Support the aging population by relieving labor gaps in healthcare; Enhance educational equity by making robotics accessible in underfunded schools; Assist workers in high-risk industries through co-bots and remote presence; Contribute to green transitions, using sustainable materials and efficient edge processing. This mission aligns with broader EU and UN targets for digital transformation, inclusion, and climate neutrality. Alba Vision is actively embedding these principles into its design philosophy: Its robots use recyclable 3D-printed parts and minimal-plastic enclosures; The company avoids centralized data harvesting, favoring on-device learning and local inference; Accessibility is built-in: from speech interaction to gesture-based control and adaptive feedback. Moreover, Alba Vision supports open innovation, planning to release portions of its stack (e.g., simulation environments, RL training configs, object models) as open-source contributions — inviting researchers, educators, and makers to co-create the future of robotics.
Chapter X – A Scalable Platform
By the end of 2025, Alba Vision’s founders envision more than just a line of humanoid robots. Their goal is to build a platform company — one that offers: Full-stack robotics kits for developers and schools; AI training environments based on real-world human-robot interaction scenarios; Remote simulation and orchestration tools for industrial robotics applications; Ethical AI and robotics consulting for institutions and enterprises. This ambition reflects their understanding that AI and robotics are not separate domains — they are converging. Success will depend on how seamlessly these technologies can integrate, adapt, and evolve with human needs. With their eyes on global standards, regulatory frameworks, and community engagement, Alba Vision stands ready to: Launch a developer beta program; Build public-private partnerships; Join the NVIDIA Inception program; Contribute to a European innovation agenda rooted in ethics, excellence, and openness.
Chapter XI – AI Autonomy Built on NVIDIA
At the heart of Alba Vision's value proposition lies the belief that true robotic intelligence emerges at the intersection of environment-aware perception, continuous learning, and safe autonomy. To realize this, the company has built its AI stack on top of NVIDIA-supported frameworks, combined with in-house research on multimodal decision systems. The AI architecture comprises four core layers: Perception and Sensor Fusion Leveraging NVIDIA Jetson Orin, stereo depth cameras, IMUs, and LiDAR systems, Alba Vision’s robots can process spatial data in real-time. Depth segmentation, visual-inertial odometry, and object tracking are handled through optimized cuDNN-enhanced CNNs, ensuring millisecond-level response times. Contextual Understanding Using transformer-based language and behavior models (e.g., fine-tuned LLMs), the robot is capable of interpreting commands, social cues, and user context. Language + visual fusion enables it to, for instance, understand “bring me the red box from the third shelf,” combining NLP, object classification, and path planning. Motion Planning and Control By integrating Isaac Sim and MoveIt into its control loop, the robot can run offline reinforcement learning for trajectory optimization and safely deploy the results through ONNX-TensorRT conversion pipelines. The RL agents are trained using curriculum learning strategies to master complex manipulation and locomotion tasks. Self-Diagnostics and Safety AI is also used for internal state prediction and failure detection, allowing Alba Vision's robots to assess joint fatigue, sensor drift, or abnormal vibrations before critical errors occur — an area where predictive maintenance meets robotic autonomy. This architecture makes the robot not just a programmable tool, but an entity capable of: Adapting to unfamiliar environments, Interacting naturally with humans, Learning from demonstrations, Operating independently at the edge. As the models evolve, so will their capability to retrain incrementally, allowing deployments to remain secure and efficient without requiring centralized data access — a core tenet of privacy-first AI.
Chapter XII – Business Model and GTM Strategy
NVIDIA is not simply a hardware supplier for Alba Vision. It is a strategic enabler. The startup is actively building on multiple NVIDIA platforms and tools: NVIDIA Tool / Platform Use Case at Alba Vision Jetson Orin / Xavier Real-time edge inference, motor control, sensor processing Isaac Sim + Isaac Lab Reinforcement learning, manipulation training, physics-based simulation Omniverse Collaborative CAD import, simulation visualization, synthetic data generation TensorRT Deployment of real-time AI models on robots CUDA / cuDNN / cuBLAS Neural network acceleration, matrix operations DeepStream SDK Object and event detection for camera-based interaction Nsight Systems Profiling and performance optimization across modules Beyond tools, NVIDIA Inception membership will unlock access to: GPU credits for training large models or simulations; Deep learning benchmarking tools; Marketing exposure; Connections with VCs, mentors, and fellow founders. In other words, Inception is not a checkbox — it is the next step in a carefully designed growth trajectory.
Chapter XIII – The Long-Term Vision
Alba Vision is entering a rapidly expanding global market for service robotics, projected to reach over $50B by 2030. Yet the company’s approach is focused on horizontal scalability and platform reuse. Revenue Streams: Modular Humanoid Platforms (hardware kits + onboard software); Robotics-as-a-Service (RaaS) subscriptions for education and retail; Licensing of simulation + AI models for industry integration; Consulting and deployment services in ethical robotics and AI. Go-to-Market Roadmap (2025–2027): Phase Objective Activities Q3 2025 MVP & Pilot Projects Deploy 3 robots in retail & 1 in education setting Q1 2026 Regional Scaling 25 units in logistics and STEM education, Romania + DACH markets Q4 2026 Developer Ecosystem SDK launch, community demos, simulation challenge 2027 Full commercialization CE-certified products, EU funding rounds, EU/US expansion Pricing will follow a modular model, where institutions or companies pay for: Base unit, Industry-specific sensors/tools, Annual AI upgrade packs, Maintenance or edge inference subscriptions. The business model is inspired by NVIDIA’s own approach to modular GPU stacks: a core platform with scalable performance via software and extension modules.
Chapter XIV – Call to Action: Let’s Build the Future
While Alba Vision’s roots are in engineering and its roadmap includes specific use cases, the underlying purpose transcends any single vertical. The founders believe in three long-term goals: Make humanoid robotics accessible to small players Most robotics systems are expensive, fragile, or locked in proprietary software. Alba Vision wants to offer affordable robots that educators, small retailers, and even NGOs can use — just like Raspberry Pi or Jetson democratized AI prototyping. Establish ethical standards in emerging AI-human interaction The company advocates for clear guidelines on data privacy, human dignity, task delegation, and algorithmic transparency. Robots must serve humanity, not replace it. Enable rapid customization through digital twins and generative AI Through Omniverse + Isaac Sim, customers will be able to simulate environments, test behaviors, and even generate task-specific AI agents using LLMs — creating tailored robots in weeks, not years. In this way, Alba Vision is part of a broader movement: the decentralized robotics revolution, where smaller actors can build meaningful AI-powered solutions without relying on Big Tech monopolies.