AI Researcher (Video, Audio, Diffusion Models)

Key Responsibilities

• Conduct deep learning research focused on video, audio, diffusion models,

transformers, and related generative or representation-learning approaches.

• Explore cutting-edge methods in multimodal modeling, temporal modeling, and

high-fidelity generation.

• Evaluate and synthesize recent research papers, identifying opportunities for novel

architectural improvements or new research directions.

• Prototype new model architectures, algorithms, and training methodologies to validate

ideas.

• Build proof-of-concept systems (POCs) to demonstrate feasibility and real-world

potential.

• Develop experiments at scale, including data preparation, model training, benchmarking,

and ablation studies.

• Collaborate with engineering teams to transition successful research into

production-ready components.

• Stay current with advancements in deep learning, GPU acceleration, and emerging AI

techniques.

Required Qualifications

• Strong understanding of deep learning theory and practical application, including

frameworks like PyTorch or JAX.

• Experience with transformer architectures, diffusion models, or generative modeling paradigms.

• Expertise in one or more of the following domains:

• Video understanding, video generation, temporal models

• Audio modeling, speech, or acoustic representation learning

• Image or multimodal generative models

• Proficiency in CUDA programming, GPU optimization, and performance engineering.

• Ability to implement research ideas cleanly and efficiently in code.

• Strong math foundation (probability, optimization, linear algebra).

Preferred Qualifications

• Publications in top-tier AI/ML conferences (NeurIPS, ICML, ICLR, CVPR, ICCV, Interspeech, etc.).

• Experience training large-scale models with distributed compute.

• Familiarity with model scaling laws, synthetic data generation, self-supervision, or curriculum learning.

• Ability to reason about architectural trade-offs, data requirements, and computational efficiency.

• Experience working with large video/audio datasets and modern preprocessing pipelines.

Experience:
Min 2 years