Models that see in real time. A solver that designs the shape.
Vayuvahana Technologies develops technology at the intersection of artificial intelligence, computational engineering and aerospace & defence. We build our own software and research platforms—from Vajra computer vision to HCFD aerodynamic design—and apply them to demanding engineering problems.
R&DTechnology developed from first principles, built for real engineering problems
Our approach is simple: research deeply, build in-house, measure honestly, and turn promising ideas into usable technology.
Current development spans real-time computer vision and aerodynamic simulation and optimisation.
Research first. Engineering all the way through.
Vayuvahana develops the core algorithms, software and systems behind its technology rather than positioning itself only as an integrator. Experimental results, benchmark data and deployment constraints guide the transition from research to engineering.
Computer vision
Real-time detection, segmentation, pose and related visual intelligence for demanding image-based applications.
Computational design
Compressible-flow simulation and adjoint-based optimisation aimed at reducing the cost and time of aerodynamic design iteration.
From research to deployment
Technology is evaluated against measurable performance, target hardware, integration constraints and the needs of the intended application.
Vajraवज्र · the thunderbolt
A family of real-time vision models developed around the YOLO lineage and re-engineered for accuracy, efficiency and deployment. One framework covers detection, segmentation, pose and more.
Open source under AGPL-3.0. Commercial licences available from Vayuvahana.
VajraV1 on COCO
| Model | mAP | Against the previous best at that scale |
|---|---|---|
| Nano | 44.3 | +3.7 over YOLOv12-N, +2.7 over YOLOv13-N, at comparable latency |
| Small | 50.4 | +2.4 over YOLOv12-S and YOLOv13-S |
| Medium | 52.7 | +0.2 over YOLOv12-M |
| Large | 53.7 | +0.3 over YOLOv13-L |
| Xlarge | 56.2 | COCO val mAP reported for VajraV1-Xlarge |
COCO validation set. Source: VajraV1 technical report, arXiv:2512.13834.
Tasks
- Object detection
- Instance segmentation
- Pose estimation
- Oriented boxes
- Classification
- Tracking
- Open-vocabulary detection
AGPL-3.0
Code and weights on GitHub for research and open projects.
github.com/NamanMakkar/VayuAIEnterprise licence
For closed-source products and on-device deployments, licensed directly by Vayuvahana Technologies.
Ask about licensingHCFDAerodynamic design software
Under developmentEngineers bring a baseline geometry and their requirements. HCFD is being developed to optimise aerodynamic shape while preserving the lift and geometric constraints that matter to the design.
A compressible flow solver and an adjoint optimiser, built as one product by Vayuvahana.
What sets it apart
Many design variables at the cost of a few simulations
An adjoint solve gives the sensitivity of drag or lift to every shape parameter at once. Refining a shape with hundreds of parameters costs about the same as tuning a handful, where a parameter sweep needs a new simulation for each one.
Gradients you can trust
The adjoint is derived from the exact equations the solver runs, and checked against brute-force simulation. An inaccurate gradient still looks plausible, which is how optimisers quietly settle on the wrong shape.
One tool from geometry to optimised shape
Shape control, mesh deformation, flow solution and the constrained optimiser work as one system. Lift and geometry requirements are enforced during the optimisation instead of being checked by hand afterwards.
Built for reproducible engineering
Reproducibility and numerical verification are part of the development process, helping engineers compare runs, audit results and iterate with confidence across development environments.
A modern, proprietary codebase
HCFD is being developed as a proprietary Rust codebase with a focus on performance, maintainability and clear licensing for commercial use and integration.
Planned multi-physics
Coupling more physics into the same adjoint lets one optimisation weigh every requirement against the others, instead of handing a design back and forth between separate tools.
Aerothermodynamics
Flow + heat transferPredict surface heating on high-speed vehicles and shape them to reduce thermal loads alongside drag.
Aero-structural design
Flow + structural deformationCouple aerodynamic loads to how the structure deflects, so a wing is designed in the shape it takes in flight and its shape and structure are optimised together.
Shape optimisation with RCS reduction
Flow + radar signatureReduce drag and radar cross-section in the same optimisation, so a low-observable design gives up as little aerodynamic performance as possible.
Built for
- Wing and aerofoil design
- UAV and aircraft programmes
- Aerospace engineering consultancies
- High-speed vehicles (planned)
HCFD is intended to move high-value aerodynamic decisions earlier in the design cycle, when geometry is still relatively inexpensive to change and optimise.
Engineering & R&D ServicesTechnical consulting and co-development for aerospace and UAV programmes
Vayuvahana works directly with engineering and technology teams on technically demanding problems across aerodynamic design, computational engineering and computer vision.
Engagements can be structured around a defined technical study, a focused prototype, or longer-term co-development alongside the customer’s engineering team.
How we work with you
A clearly defined problem
We take on a specific engineering or R&D question—such as aerodynamic analysis, shape optimisation, design review, computer-vision feasibility, or model development—and deliver a defined set of technical results, documentation and recommendations.
An extension of your engineering team
We work alongside your programme from problem definition through prototype development, taking responsibility for agreed technical work packages and integrating our models, software or analysis into the wider system.
Technical capabilities
Our services are built around the same technologies we develop in-house, from aerodynamic optimisation and simulation to computer vision and deployment.
Aerodynamic design & CFD
Aerofoils · wings · airframesAerodynamic analysis, performance assessment, drag reduction and adjoint-based shape optimisation for aerofoils, wings and aircraft or UAV geometries.
Computer vision & perception
Detection · segmentation · trackingCustom vision models for aerial and other mission-specific imagery, using Vajra as a development foundation where appropriate and tailoring the system to the customer’s data and operating requirements.
Deployment & performance engineering
TensorRT · edge GPUs · benchmarkingModel export, integration and performance benchmarking on the target compute platform, with emphasis on measured inference latency and deployment constraints rather than theoretical benchmarks alone.
Work with Vayuvahana Technologies
Tell us what you are building, the technical requirement you are trying to solve, or the capability you want to develop.
Prefer email? namanmakkar@vayuvahanatechnologies.com