Realtime Camera-and-Mic Semantic Pipeline

The build sends camera and microphone input through Moonshine, YOLO26, and DepthART, with Jev adding a semantic layer. It uses ruNNtime for local inference and TypeGPU to share GPU resources between inference and rendering without copying.

View on X

Konrad Reczko@reczko_konrad𝕏

TypeGPU + ruNNtime + Jev @typesafeai is a very fun combo :D ruNNtime gives me efficient local inference, TypeGPU lets inference and rendering share GPU resources directly with zero copy. That’s 3 separate NN inferences plus rendering, all happening in realtime Since we control the pipeline, Jev can just sit in the middle and add the semantic bit. camera + mic → Moonshine + YOLO26 + DepthART → Jev

Sep 17, 2026X postsView on X

The author says three separate neural-network inferences and rendering happen in real time. They also note that they control the pipeline.

Also filed under Games & real time

Steer a 3D Box with Goals
The build steers a 3D box with plain-language goals. An LLM plans each instruction, while TypeSafe Jev makes move, turn, and jump decisions in real time.
Benchmarks Jev on Pokémon Red
The project benchmarks Jev, TypeSafe’s fast decision model, against Jev paired with a GPT-6 Sol planner for playing Pokémon Red.
A Jev-Made Unity Build
Kevin Kern says the build was made with Jev, GPT Live, Blender, Unity, and more.
Plain-English Cellular Ecosystem Simulator
LifePot is a cellular automaton-inspired ecosystem that users describe in plain English. Jev turns those inputs into species, feeding relationships, reproduction strategies, and environmental conditions.