About this role
Own large-scale speech/audio data pipelines and support fine-tuning and evaluation of speech/language models (SpeechLLM or general LLM), focusing on improving recognition accuracy for code-switching, names, and product terms; perform domain adaptation and build evaluation frameworks to benchmark internal models against baselines.
Skills for this role
Automatic Speech Recognition (ASR)SpeechLLMLLMMachine LearningData EngineeringAudio Data PipelinesLarge-scale Data ProcessingData CollectionData CleaningData LabelingData AugmentationQuality ControlTerm/Hotword MiningDomain AdaptationModel Fine-tuningModel EvaluationEvaluation FrameworksTest Set CreationBenchmarkingPythonPyTorchDistributed Data ProcessingSparkRayHotword biasingContextual biasingCode-switching ASRStepAudioQwen3-OmniMultimodal Data