Tavus
Multimodal AI Model Optimization Research Engineer
About this role
Tavus seeks a Research Scientist/Engineer to optimize multimodal AI models for production deployment. You'll apply advanced compression and efficiency techniques to make cutting-edge models fast, cost-effective, and ready for real-time applications in healthcare, sales, and education.
What you'll do
- Optimize research models using sparsification, distillation, quantization, and mixed precision techniques
- Own the full optimization lifecycle: define metrics, run experiments, benchmark latency/cost/quality trade-offs
- Partner with researchers and engineers to translate new ideas into deployable systems
- Profile and tune inference performance on GPUs and accelerators
- Reproduce ML papers and adapt optimization techniques to multimodal (audio/video/language) models
- Track experiments and maintain benchmarking standards at scale
What they're looking for
- PyTorch and deep learning
- Model compression and optimization (distillation, pruning, quantization)
- Inference performance and GPU/accelerator fundamentals
- Python and research engineering best practices
- Large-scale model and dataset handling in cloud environments
- ML paper reproduction and adaptation
- Collaboration and technical communication
- Profiling and benchmarking tools
Benefits
- Flexible work schedules and unlimited PTO
- Competitive healthcare and gear stipends
- Collaborative, learning-focused environment
- Remote-friendly or hybrid SF options with relocation support
- Series B funding and backing from top-tier investors
- Opportunity to impact human-AI interaction at scale
Opens the application — the Jobs AI extension fills it for you. Set up autofill
Opens the official application on the employer’s site. No login required.
Tavus
Tavus builds AI-driven avatar technology powered by foundation multimodal conversational models that enable real-time verbal and non-verbal interactions. The company is hiring research engineers, data engineers, infrastructure specialists, and customer-facing engineers to advance its conversational AI capabilities and improve developer experience.
- Website
- tavus.io
Likely interview questions
- Walk us through a recent model optimization project—which technique (distillation, quantization, pruning) had the biggest impact on latency and why?
- How do you approach the trade-off between model quality and inference speed? What metrics do you prioritize?