Location: Paris preferred (remote within France/EU possible)

Duration: 6 months (stage de fin d’études)

Team: Research

Start: Flexible (earliest: September)

AudioShake is looking for a Master’s-level intern to join our research team for a 6-month internship focused on model optimization. You’ll help to scale our deep learning based speech and music models for deployment — both in the cloud and on edge devices with the focus on real-time and low latency.

This role is hybrid-friendly, ideally based in Paris (where part of our team is located), but we also welcome remote applicants from elsewhere in France or Europe.

The role also offers the possibility of conversion to a full-time position after the internship.

💼 What we offer

• An opportunity to work with state-of-the-art audio technologies
• An empowering and fast-paced working environment.
• International team of excellent researchers, well connected to academia and industry

🎯 What You’ll Do

• Dive deep into the internals of our PyTorch models
• Identify bottlenecks — and remove them
• Profile, prune, quantize, and re-structure model components
• Benchmark exports and compare across optimization strategies
• Collaborate with researchers to test architecture variants
• Help prepare models for cloud and edge deployment
• Collaborate with a team of researchers and engineers to bring prototypes into production

🤹 Requirements

• Master’s student in CS, engineering, or applied math (end-of-studies internship)
• Strong Python & PyTorch skills — you know what’s under the hood of a Conv1d
• Passionate about clean, maintainable & efficient code
• Mastery of Python in Linux & Unix environments

⭐ Nice-to-Have

• Experience with ONNX, torch.script, torch.compile, quantization, or pruning tools
• Familiar with profiling tools like torch.profiler, PyTorch benchmarks, or similar
• Basic understanding of audio signals or DSP (not required)
• Interested in performance engineering and model internals