Work on expressive speech, voice agents, and multimodal AI at Boson AI in Toronto, using Python, PyTorch, or JAX on a hybrid schedule.

As an MScAC Applied Research Intern, you will contribute to projects supporting Higgs models, including Higgs Realtime, Higgs TTS 3, and Higgs Avatar. Research may involve multilingual audio, conversational evaluation, video generation, reinforcement learning, or full-duplex interaction. The internship begins in May 2027.

Key Responsibilities:

- Evaluate expressive, multilingual, and conversational speech models
- Automate evaluation using Python and measurable benchmarks
- Work on audio-driven video generation and accurate lip sync
- Develop agents for workflows, user modeling, and call traces
- Model interruptions, backchannels, and real-time turn-taking

Requirements:

- Mandatory University of Toronto MScAC enrollment
- Solid programming skills in Python
- Solid machine learning and deep learning foundations
- Hands-on PyTorch or JAX experience
- Careful experimental design and honest reporting

Bring research experience in speech, audio, video, agents, or multimodal learning to Boson AI's applied research projects.

📌 Speech and Multimodal AI Intern (Toronto)
🏢 Boson AI
📍 Toronto