About Maple Silicon

Built by a GPU systems engineer for teams who pay real money for inference.

Maple Silicon Inc. is a Canadian company building systems for AI inference acceleration and passive drone detection in real-world operating environments.

Most optimization stories sound compelling until they meet real power, cost, and latency constraints. Maple Silicon is built around making advanced computation practical under those constraints instead of pretending they do not exist.

Maple Silicon Inc., incorporated in Canada A100 validated; RTX 4090 benchmarking in progress Open-source SparseFlow repo available on GitHub Founder-led product conversations and pilot evaluations
Gourav Kumar
Founder & Lead Engineer

Background in CUDA kernel development and GPU systems engineering. Built dense GEMM kernels reaching 31.47 TFLOPS on RTX 3090. Currently leading Maple Silicon product development and customer engagements end-to-end.

"The next performance win is not just a better model. It's better execution on the GPUs teams already pay for."

Dual-System Approach

Two complementary systems, built on one engineering bias.

SparseFlow

Compiler-driven inference engine

A compiler-driven inference engine focused on accelerating machine learning workloads efficiently across GPU architectures.

Maple Shield

Passive drone detection system

A passive drone detection system designed for contested and remote environments, powered by the CAIRN detection engine and built for zero-RF passive coverage.

Shared foundation

Power, cost, and latency first

Both systems are built on a shared foundation: making advanced computation practical under real-world constraints - power, cost, and latency.

Proof And Updates

What a technical buyer can verify quickly

Public Repo

SparseFlow on GitHub

Review the public product surface, benchmark framing, and current repository for SparseFlow.

Benchmark Framing

Results before claims

See how SparseFlow presents speedup, workload fit, and evaluation scope before a pilot conversation starts.

Commercial Path

Free review, then pilot

Start with a lightweight benchmark review, then move into a paid pilot when your workload shows real upside.

Maple Shield was submitted to NATO's Innovation Continuum 2026 for evaluation under the Layered Counter-UAS Initiative.