About the role#
The Multimedia Lab at ByteDance works on video understanding, perceptual quality, and intelligent coding. We build systems that balance quality, efficiency, and speed. As a PhD intern, you will bridge the gap between research and production. You will work with researchers and engineers to see how your ideas affect the cost and quality of video across our products. This role focuses on how multimodal large language models can understand video content, reason over time, and assess quality.
What you'll need#
- Currently pursuing a PhD in Computer Science, Computer Engineering, Electrical Engineering, or a related field.
- Strong foundation in machine learning and computer vision.
- Experience in multimodal learning, video understanding, or video processing.
- Proficiency in Python and experience with frameworks like PyTorch or TensorFlow.
- Ability to design experiments, implement research, and communicate technical findings.
- Familiarity with Transformer architectures, vision-language models, or MLLMs.
- Hands-on experience with model training, fine-tuning, and evaluation.
- Preferred: Experience with long-video understanding, perceptual modeling, video compression, or diffusion models.
- Preferred: Familiarity with Linux, OpenCV, FFmpeg, or C/C++.
- Evidence of research capability through publications or substantive projects.
What you'll do#
- Research and prototype multimodal models for video tasks like temporal reasoning, event localization, and salient object detection.
- Develop methods to identify and explain spatial and temporal video degradations.
- Study how user preferences and semantic context influence perceived quality.
- Build datasets and benchmarks, including designing annotation protocols and analyzing model robustness.
- Explore efficient inference for multimodal foundation models, such as parameter-efficient adaptation and visual token compression.
- Connect video understanding with enhancement and encoding strategies through offline experiments and production A/B testing.
Location & details#
- Location: San Jose, California.
- Modality: On-site.
- Employment: Full-time internship.
- Compensation: $60 per hour.
- Timeline: Rolling basis.


