Frequently Asked Questions

No. We provide a lifecycle digital twin — design model + inspection condition + operational and emissions data — not a live sensor-fed twin. We don't claim real-time IoT fusion or autonomously predictive simulation; we provide a single, coherent model spanning design, inspection, and operations of the vessel.

Yes. Navatala's inspection apps are offline-first. You can conduct inspections, capture photos, and record deficiencies without connectivity. Data syncs automatically when you reconnect using delta synchronization.

EU FuelEU Maritime, IMO Data Collection System (DCS), Carbon Intensity Indicator (CII) reporting, and classification-society requirements. The system tracks fuel consumption, emissions, and generates compliance documentation.

navatala_gpu is a multi-backend GPU compute runtime and an AI/scientific-computing kernel corpus, released under the Apache License 2.0 as a developer preview. It supports CUDA, HIP/ROCm, OpenCL, Vulkan compute, and Metal, with kernels for CFD primitives, sparse linear algebra, vector search, DNN building blocks, dataframe operations, and ML algorithms.

Repository: github.com/navatala-systems/navatala_gpu. Install: pip install navatala-gpu.

CUDA, HIP/ROCm, OpenCL, Vulkan compute (+ SPIR-V), and Metal. Coverage is not uniform — CUDA and HIP/ROCm currently have the broadest generated coverage; OpenCL, Vulkan, and Metal have platform-specific limitations.

See the live coverage matrix at docs/BACKEND_COVERAGE.md.

Yes. Hand-authored layers (runtime, examples, tests, docs, tooling) accept normal pull requests. Generated paths (kernel sources, Python facade modules) route bug reports and reproducers through the maintainers, who apply the fix internally and regenerate the public tree. The repository ships a Regen-Manifest-Trailer: git hook that enforces the distinction.

See CONTRIBUTING.md for the full contribution model.