Latest AI JobsBrowse Jobs
Tavus logo

Multimodal AI Model Optimization Research Engineer

TavusRemoteEngineering, Product, & Design
Apply on company site
Salary
Not listed
Job type
Full-time
Work mode
Remote
Posted
28 Sept 2026
Status
Open · checked 7d ago

About this role

Tavus is hiring a remote Multimodal AI Model Optimization Research Engineer to build fast, efficient, and production-ready deep learning models. This full-time position suits an experienced research scientist who wants to drive the future of natural human-AI interaction in a fast-paced startup environment.

Role Overview

The core AI team at Tavus needs a dedicated engineer to transform cutting-edge research models into highly efficient, production-ready systems. This professional will focus on bridging the gap between theoretical machine learning breakthroughs and real-world deployment by maximizing inference speed and resource efficiency.

Key Responsibilities

Transform advanced research models into fast, efficient, and production-ready applications through quantization, distillation, and sparsification.
Oversee the complete optimization lifecycle for critical models by establishing metrics, conducting experiments, and evaluating trade-offs between latency, quality, and cost.
Collaborate closely with cross-functional engineering and research colleagues to translate innovative concepts into robust, deployable systems.

Requirements

Extensive background in deep learning utilizing PyTorch.
Practical expertise in model optimization and compression techniques such as mixed precision, pruning, and knowledge distillation.
Knowledge of efficient neural network architectures, including low-rank adapters.
Deep familiarity with inference performance metrics and fundamental GPU or accelerator operations.
Advanced programming capabilities in Python paired with sound research engineering practices.
Background handling massive models and datasets within cloud infrastructure.
Capacity to digest machine learning literature, replicate findings, and modify concepts effectively.
Strong verbal and written communication skills to foster teamwork.

Nice to Have

Background in optimizing large language models, diffusion models, or audio-visual generative systems.
Exposure to streaming or real-time architectures, including low-latency APIs, WebRTC, and streaming video or text-to-speech.
Knowledge of deployment and acceleration tools like TensorRT, ONNX Runtime, TVM, Triton, or XLA.
Proficiency in writing low-level performance tuning or custom CUDA and Triton kernels.
Track record in profiling, benchmarking, and tracking experiments at scale.
Previous background working in applied science or research engineering capacities.

Skills and Tools

Python
PyTorch
Model compression and optimization (quantization, distillation, pruning)
Cloud environments and large dataset handling
GPU and accelerator infrastructure

Salary and Benefits

Flexible working schedules
Unlimited paid time off
Competitive healthcare coverage
Gear stipends
Collaborative, impact-driven work culture
Relocation support for candidates preferring a hybrid setup in San Francisco

About the Company

Tavus develops pioneering multimodal AI technology designed to model human-to-human communication across language, audio, and video. Backed by prominent investors including Sequoia, Y Combinator, and Scale VC, the organization creates natural conversational video experiences for sectors like education, healthcare, and sales.

How to Apply

Interested candidates should visit the official Tavus careers page to submit an application for this remote position.

Skills and tech stack

PythonPyTorchDeep LearningCUDADiffusion

Questions candidates ask

Is this position fully remote?

Yes, remote candidates are considered, though the position is also available as a hybrid role in San Francisco with relocation support.

What primary framework is required for this role?

Candidates must have strong experience in deep learning using PyTorch.

What are the main optimization techniques involved in the job?

The role involves model optimization and compression techniques including knowledge distillation, pruning, sparsification, quantization, and mixed precision.

What kind of benefits does the company provide?

Benefits include flexible work schedules, unlimited paid time off, competitive healthcare, gear stipends, and relocation support if choosing the San Francisco hybrid option.

Who backs Tavus?

Tavus is a Series B company backed by top investors including Sequoia, Y Combinator, and Scale VC.

Ready to apply?

The application is on the official Tavus careers page.

Apply on company site