I read molecules by the light they scatter. My work sits where surface-enhanced Raman spectroscopy, plasmonic nanostructures and machine learning meet — building optical platforms that identify a pathogen or a biomarker in minutes, from a signal one photon in a billion strong.
Centre for Photonics and Quantum Communication Technology · Indian Institute of Technology Roorkee · research group of Prof. Sachin Kumar Srivastava.
About
I am a Senior Research Fellow (SRF) at the Centre for Photonics and Quantum Communication Technology (CPQCT), IIT Roorkee, working with Prof. Sachin Kumar Srivastava (Department of Physics and CPQCT). My subject is what light gives away about matter — specifically, the vanishingly small fraction of scattered photons that come back carrying a molecule's vibrational signature.
Most of my time goes to surface-enhanced Raman spectroscopy: designing plasmonic substrates that amplify that signature by orders of magnitude, and then working out how much of the result is the sample and how much is the substrate. The current application is a receptor-free platform that identifies respiratory viruses from their intrinsic spectrum, with machine learning doing the separation that no single band can.
The rest goes to the things that make the first part usable. I wrote RNSpectraX because a reproducible analysis pipeline should not require three programs and two undocumented settings. I record Lab2Life because a result nobody outside the field can follow is only half finished.
What I work on
How little of a molecule can you detect, how fast, and how certain can you be about what it is? Each thread below attacks a different part of that question — the surface, the signal, the platform, and the analysis.
Raman scattering is famously weak — roughly one photon in a billion. Park a molecule nanometres from a plasmonic surface and that signal jumps by orders of magnitude. I build substrates that hold analytes in those hot spots reproducibly, so the enhancement is a measurement rather than an accident, and detect targets by their intrinsic fingerprint instead of a labelled receptor.
A biological SERS spectrum is a crowded superposition: hundreds of overlapping bands, a drifting fluorescence background, and sample-to-sample variation that dwarfs the difference you care about. Classifiers trained on properly preprocessed spectra separate classes that no single band distinguishes — provided the preprocessing is honest about what it removes.
A spectrometer on an optical bench is not a diagnostic. Fibre Bragg gratings, waveguide couplers and integrated optical elements move a measurement from the bench toward something field-deployable, and each introduces its own trade-off between sensitivity, robustness and cost.
Two labs applying "baseline correction" to the same spectrum can publish different peak ratios. RNSpectraX exposes every processing choice — despiking threshold, baseline order, smoothing window — so a pipeline can be reported precisely and reproduced exactly. It runs entirely in the browser, so no data leaves the machine that measured it.
Software
A full Raman and SERS analysis workspace that runs in a browser tab. Built because moving spectra between three pieces of proprietary software to make one figure is not a workflow — it is a reproducibility problem.
How to cite (APA and BibTeX) · Cite as: Naik, R., & Srivastava, S. K. (2026). RNSpectraX (Spectral Plotter and Analytics Suite) (Version 3.4) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.23593680
A real RNSpectraX result: ALS baseline correction, group averages and automatic peak labels (synthetic demonstration spectra).
Software
An optical setup designer for publication figures and graphical abstracts. You place the optics; the light is traced for you, so the figure is also a working model of the experiment.
By Rakesh Naik and Sachin Kumar Srivastava, IIT Roorkee. Runs entirely in the browser; nothing is uploaded. · How to cite (APA and BibTeX) · Cite as: Naik, R., & Srivastava, S. K. (2026). RNBeamX (Optical Setup Designer) (Version 1.2) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.23697703 · Works hand in hand with RNSpectraX: send a processed spectrum straight across as panel b.
Exported from RNBeamX as vector SVG: every beam is traced by the app.
Rendered by RNBeamX: the 3D view on a brushed stainless-steel table, a turntable swing and light flow.
How it has gone so far
Started advanced work on Raman spectroscopy and optical sensing at the Centre for Photonics and Quantum Communication Technology, and began pairing spectral measurement with machine learning.
Developed a machine-learning-integrated SERS platform for rapid respiratory virus detection — identifying targets from their intrinsic vibrational fingerprint rather than from a bound receptor, which removes the reagent that is usually the slowest and most fragile part of a diagnostic.
Released RNSpectraX (Spectral Plotter and Analytics Suite), presented at the OPTCT + SeNcity conference at IIT Roorkee, and started publishing science communication — explainers and lab footage — under Lab2Life.
Beyond the bench
A result that nobody outside the field can follow is only half finished. These are the places I try to close that gap.
A video channel on Raman scattering, SERS, photonics and AI in healthcare — explainers, lab demonstrations and the parts of research that never reach a paper.
WatchShort pieces on research, campus life at IIT Roorkee, and the occasional encounter with the local wildlife that ends with a lesson in colony defence.
Read the postsConferences, demonstrations, group milestones, and a stretch of Japan between Matsue Castle and Lake Shinji.
Look aroundGet in touch
I am glad to hear from anyone working on SERS substrates, spectral classification or optical diagnostics — and from students thinking about a PhD in photonics.