AI Research Intern

About the role

Join our engineering team to advance our work in multimodal AI, LLMs, computer vision, speech technology, and agentic systems. You will contribute directly to the OpusClip and AgentOpus platforms by building prototypes and deploying features for our global user base.

Key facts

What you'll do

• Develop deep learning models focused on computer vision, audio processing, or multimodal generation.
• Integrate foundation models into production using RAG frameworks, prompt engineering, and agent workflows.
• Create and maintain scalable evaluation pipelines for multimodal systems.
• Build domain-specific benchmarks using automated metrics and LLM-as-a-Judge techniques.
• Translate recent academic research into functional, production-ready code.
• Collaborate with engineers to prototype and launch new capabilities across our product suite.

Requirements

• Currently enrolled in or recently finished a Master's degree in Computer Science, AI, Mathematics, or a related field.
• Strong grasp of Transformer architectures, attention mechanisms, and generative model families like diffusion, GANs, and autoregressive models.
• Proficiency in Python, Linux, Git, and data structures.
• Experience with media processing fundamentals including codecs, ffmpeg, or signal processing.
• Ability to read technical papers and produce clear documentation in English.
• Hands-on experience in at least one area: low-level computer vision, voice cloning or enhancement, or LLM fine-tuning techniques such as SFT, RLHF, DPO, or LoRA.

Nice to have

• Experience building agent systems or features using APIs like Gemini, Claude, or ChatGPT.
• Proficiency in TypeScript.
• A track record of end-to-end project ownership or open-source contributions.
• Academic background in video understanding, multimodal systems, or model benchmarking.
• Publications in venues such as CVPR, ICCV, NeurIPS, ICML, ICLR, ICASSP, Interspeech, or ACL.

Skills & tools

• Python, TypeScript, Linux, Git
• LangChain, RAG frameworks
• Transformer architectures, GANs, Diffusion models
• ffmpeg, codecs, signal processing
• SFT, RLHF, DPO, LoRA, PEFT

Practical notes

• This is an on-site role in Mountain View.
• Attendance is required 3 days per week, with 4+ days per week preferred.