MSc Internship/thesis project: Towards Quantitative SERS: Where Raman Spectroscopy Meets AI

MSc Internship/thesis project: Towards Quantitative SERS: Where Raman Spectroscopy Meets AI 📍 Location: Amsterdam 🕒 Timing: within two months, ~6–9 months (flexible) 💼 32–40 hours/week About the project Nostics is developing a next-generation diagnostic platform for the rapid detection and identification of bacterial infections. Our technology uses Surface-Enhanced Raman Spectroscopy (SERS) to read molecular fingerprints and support faster, better-informed clinical decisions. SERS is exceptionally sensitive, but unlike Raman, its signals are not always easy to predict or quantify. When a molecule interacts with a SERS surface, some peaks become stronger, others shift or disappear, and the resulting spectrum can look very different from a conventional Raman spectrum. In this project, you will combine experimental spectroscopy, data analysis, and machine learning to unravel that transformation. You will create a controlled dataset of paired Raman and SERS measurements and develop models that predict how a molecule’s Raman fingerprint changes on the Nostics substrate. The ambition: take a meaningful step towards predictable, reproducible, and ultimately quantitative SERS . Research question Can the relationship between Raman and SERS spectra be modelled well enough to make SERS responses quantitatively predictable on the Nostics substrate; and where does that predictability break down? What you’ll do You’ll own a self-contained research question spanning the full loop from measurement to model: Develop a standardized, reproducible protocol to measure both the SERS- and Raman spectra of selected compounds. Measure relevant analytes using SERS and Raman spectroscopy to build a paired spectral dataset of traceable compounds. Characterize how SERS systematically differs from Raman across your sample set (which bands enhance, shift, or disappear). Design and validate a model that predicts SERS spectra from Raman spectra, from a physically interpretable enhancement factor per band to a machine-learned mapping. Map the domain of validity of that model — where the mapping holds and where it breaks down. Document your protocol, dataset, and results for internal use and future research.

Enrolled in an MSc in Analytical Chemistry, Nanoscience, (Bio)Physics, life sciences or a data-driven fields with a solid chemistry background and lab experience. Rigorous in the lab and comfortable analyzing data in Python (e.g. NumPy, pandas, scikit-learn). Strong analytical thinking and the ability to structure and communicate results clearly. Interested in the physics of why SERS behaves the way it does, not just in running a model.

Experience with spectral data, signal processing, or chemometrics. Experience building reproducible data pipelines. Familiarity with deep learning frameworks (e.g. PyTorch, TensorFlow). Experience programming with the help of LLM-based coding tools (e.g. Claude Code, Codex). What you’ll learn How to design and run a rigorous spectroscopy experiment from protocol to publishable dataset. The physics and chemistry behind SERS enhancement, and how to model a non-linear physical relationship. How to combine experimental work with data-driven modelling in a real diagnostic product. How interdisciplinary teams (biology, chemistry, physics, engineering, data science) collaborate. How early-stage diagnostic technologies are developed and validated.

Mentorship from an excellent multi-disc i plinary team applying spectroscopy and machine learning in a real diagnostic product; and hands-on experience across the full loop from measurement to model. Strong results may contribute to a publication. You will also receive standard reimbursement for your internship.