Member of Technical Staff, NeuroAI

Neurosoft Bioelectronics is a neurotechnology startup that builds next‑generation AI for decoding high‑density subdural ECoG LFP time series. The team is looking for an intern machine‑learning scientist who will help create foundation models that translate a few minutes of brain recordings into precise finger movements and full‑hand kinematics. On a daily basis you will design and benchmark sequence‑model architectures such as Kalman filters, RNNs, LSTMs, Structured State‑Space Models (S4/S5), Mamba and Neural ODEs. You will also develop large‑scale self‑supervised pre‑training pipelines, fine‑tune them for individual subjects, and extend state‑of‑the‑art neural‑decoding methods. The work includes decoding individual finger motions, grasp synergies, continuous gestures and full 6‑DoF hand pose, as well as deploying low‑latency streaming inference with ONNX, TensorRT, Triton and GPU acceleration. Maintaining reproducible benchmarking, MLOps and the research‑production infrastructure is part of the role. The position requires solid familiarity with Python, C/C++, Java, MATLAB, LaTeX, HTML, JavaScript, PyTorch and TensorFlow, together with a good grounding in signal‑processing and machine‑learning concepts. You should be capable of conducting independent research, writing scientific reports, planning strategically, managing time effectively and presenting results clearly. Additional, optional strengths are welcomed. A theoretical background in time‑series modelling, system identification, dynamical systems, differential equations, control theory or Bayesian estimation is valuable. Prior implementation experience with Kalman filters, HMMs, linear/non‑linear SSMs, RNNs, LSTMs, S4/S5, Mamba, Neural ODEs or transformer‑based sequence models is a plus, as is exposure to contrastive learning, VAEs, foundation‑model pre‑training, transfer learning, domain adaptation or few‑shot learning. Proficiency with Python‑centric tools such as CUDA, Docker, ONNX, TensorRT, Triton, GitHub, Linux and MLflow further strengthens your profile. The role offers a hybrid work model based in Geneva, direct interaction with founders, an entry‑level salary of 45 000‑55 000 CHF, a quarterly‑reviewed CSO‑track growth path and possible equity participation. You will join a collaborative, dynamic team and gain exposure to multiple facets of a fast‑growing neurotechnology company. What the role asks for: - Familiarity with Python - Familiarity with C/C++ - Familiarity with Signal Processing - Familiarity with Machine Learning - Independent research capability - Scientific writing capability - Nice-to-have: Theoretical foundations of time‑series modelling - Nice-to-have: Recent implementation of Kalman filters or similar