Intern, NeuroAI: Real-Time Neural Decoding

Neurosoft Bioelectronics SA; Genf; Hybrid; CHF 45,000 - 55,000; Full time

Benefits offered by this job

Hybrid work model

Job summary

Neurosoft Bioelectronics SA seeks an intern machine learning scientist to explore sequence modelling, state-space methods, and self-supervised learning for neural decoding. The role includes benchmarking Kalman filters, RNNs, LSTMs, S4/S5, and deploying low-latency inference pipelines.

Responsibilities cover data-driven research, pretraining, fine-tuning, and production-ready ML workflows, with hybrid work options in Geneva and equity consideration where appropriate.

Qualifications

• Familiarity with Signal Processing and Machine Learning.
• Independent research, scientific writing, planning, and time management.
• Optional: time-series modelling, dynamics, Bayesian estimation foundations.

Responsibilities

• Design and benchmark sequence models: Kalman filters, RNNs, LSTMs, S4/S5, Mamba, Neural ODEs.
• Develop self-supervised pretraining pipelines and subject-specific fine-tuning workflows.
• Reproduce and extend SoTA neural decoding methods from LSTM baselines to modern SSMs.
• Decode finger movements, 6 DoF hand pose, and related gestures.
• Deploy streaming inference using ONNX, TensorRT, Triton, and GPU acceleration.
• Maintain reproducible benchmarking, MLOps, and production research infra.

Skills

Machine Learning

Signal Processing

Python

PyTorch

Linux

Docker

ONNX

TensorRT

Triton

GitHub

CUDA

MLflow

Tools

CUDA

Docker

ONNX

TensorRT

Triton

GitHub

Linux

MLflow

Job description

Neurosoft Bioelectronics SA seeks an intern machine learning scientist to explore sequence modelling, state-space methods, and self-supervised learning for neural decoding. The role includes benchmarking Kalman filters, RNNs, LSTMs, S4/S5, and deploying low-latency inference pipelines.

Responsibilities cover data-driven research, pretraining, fine-tuning, and production-ready ML workflows, with hybrid work options in Geneva and equity consideration where appropriate.