MVAMulti-ModalityInteractionDeveloper

Key Responsibilities Develop based on the current mainstream speech systems , including SSPE, wakeup, vad, asr, nlu, dm, tts, LLM, and etc. Design and implement multimodal fusion combining speech, DMS camera, OMS camera, Dash camera, microphone, sensors, audio system state, voice print, and vehicle state data. Normalize and structure multimodal inputs into system context representations suitable for LLM reasoning to support future LLM-based assistant use cases, such as; context-aware dialogue, assistant memory collection and apply , and etc. Design and maintain consistent multimodal data pipelines , handling time alignment, normalization, and state coherence as data flows from vehicle systems into LLM-ready context representations. Consume vehicle system capabilities through service-oriented APIs , enabling intent-driven control of vehicle functions. Integrate and abstract data from multiple vehicle ECUs (audio, cameras, sensors, body, ADAS, etc.), with the ability to independently explore and onboard new data sources. Collaborate closely with EE, platform, AI, and UX teams, acting as a cross-team technical bridge . Required Qualifications Experience developing speech or voice assistant systems , including wake word, VAD, ASR, NLU, dialogue management, TTS, and LLM integration. Hands-on experience with multimodal data integration and fusion , combining audio, camera, sensor, and vehicle state information. Strong understanding of multimodal data pipelines , including normalization, temporal alignment, and state consistency for LLM-ready context. Practical experience using LLMs as a reasoning layer , including context preparation and safe application of outputs. Ability to consume service-oriented vehicle APIs for intent-driven control of vehicle capabilities. Experience integrating with embedded or automotive systems , working across multiple ECUs (audio, camera, sensor, body, ADAS). Solid understanding of Android system architecture , preferably Android Automotive O