Help set the roadmap for new model capabilities, short- and long-term.
Propose new multimodal architectures, especially for text and voice.
Build and evaluate streaming conversational systems for low-latency voice-video synthesis.
Design approaches that make interaction more emotionally expressive and natural.
Implement designs across pretraining and post-training.
Integrate and test neural codecs, diffusion, and flow-matching methods.
Define new conversational evaluation metrics, including latency-aware and interaction-based ones.
Track research in audio-visual diffusion, autoregressive models, codecs, and multimodal LLMs.
Curate new datasets to complement existing ones.
Lead post-training work such as DPO, fine-tuning, and distillation.
Ship models to production with optimized runtimes and act on customer feedback.

Original ideas that advance interactive multimodal systems.
Strong understanding of generative modeling, ideally on sequential or multimodal data.
Hands-on experience with LLMs or other transformer-based architectures.
High proficiency in PyTorch, including distributed training and model optimization.
Solid grasp of time-series modeling and tokenization, preferably for audio, speech, or video.
Ability to prototype quickly, test hypotheses, and iterate.
Experience training deep learning models end to end.
Strong general software engineering skills.
Bonus: experience with real-time or streaming architectures, audio and speech generation, and publications at top-tier venues.

Done for you

We'll apply to this for you

Denidel's HR team submits and tracks the application on your behalf.

Opens the employer's own site, where you complete your application

Posted	17 Oct 2026
Views	0

Staff Research Engineer - Multimodal Generative Modelling