Projects & Industry Applications

Challenge

Characterizing the polarization state at each focus of a Twin-Foci ultrafast laser setup required manually rotating a motorized waveplate, reading power at each angle, and fitting data in a spreadsheet. The process took over 2 hours per dataset and was error-prone.

Solution

Built a real-time PyQt5 platform that drives a Newport ESP301 motion controller and Ophir NOVAII power meter simultaneously, collecting intensity vs. angle data in real time. Uses the Fourier-based Schaefer method to extract complete Stokes parameters (S₀, S₁, S₂, S₃). Includes a TDC synchronization module for photon-counting experiments, an offline analysis CLI, and a hardware simulation mode for testing without physical equipment.

Impact

• •Reduced scan time from 2+ hours to ~5 minutes (96% faster)
• •Full Stokes vector extraction with uncertainty propagation from covariance matrix
• •TDC polarization controller for synchronized photon-counting experiments
• •Hardware simulation mode and offline CLI for testing and post-processing
• •Publication-quality export at 300 DPI (PNG, PDF, SVG)

Challenge

Industrial turbine engines degrade gradually across hundreds of sensor channels. Traditional monitoring catches failures too late, after damage is done. The challenge: predict remaining useful life (RUL) from noisy, high-dimensional time-series data and explain why the model raised an alert.

Solution

Built an end-to-end ML pipeline on the NASA C-MAPSS turbofan degradation dataset. The system ingests raw multi-sensor data, engineers time-series features (rolling statistics, degradation trends), trains and compares models (Random Forest, XGBoost), tracks experiments with MLflow, and serves predictions via a FastAPI endpoint. A SHAP explainability layer translates model outputs into sensor-level diagnostic reasoning.

Impact

• •Predicts engine failure window with interpretable confidence
• •SHAP layer identifies top contributing sensors per failure event
• •Full MLOps structure: experiment tracking, model versioning, API deployment
• •Streamlit dashboard surfaces health status and alerts in real time

Challenge

Time-resolved PEPICO experiments produce complex binary data from TDC acquisition. Manual conversion and analysis limited throughput and introduced inconsistencies.

Solution

Built a Python/Jupyter-based pipeline that converts raw TDC binary files to calibrated time-of-flight spectra, performs electron kinetic energy and mass calibration, coincidence analysis (TOF-vs-TOF), and statistical evaluation. Users specify data location and basic parameters; the pipeline handles the rest.

Impact

• •Processes multiple datasets reliably and consistently
• •Generates publication-ready plots and statistical summaries
• •Removes analysis as a bottleneck, experiment pace now drives progress

Challenge

Manage a prestigious national fellowship (one of ~1,000 awarded annually in India) as principal researcher while completing PhD research.

Solution

End-to-end program delivery including annual financial reporting, procurement planning, progress tracking for government review, and multi-institution collaboration across Germany and India over 5 years.

Impact

• •Delivered 12 first-author publications
• •Contributed to one granted patent (gas-separation membranes)
• •Maintained regulatory compliance across 5-year program
• •Coordinated with stakeholders across multiple institutions

Challenge

In a shared ultrafast laser laboratory, scheduling conflicts, documentation gaps, and workflow handoffs limited overall productivity.

Solution

Team-based implementation of shared equipment booking calendar, standardized maintenance checklists, experimental protocols, and Obsidian-based shared lab notebook. Weekly planning meetings align schedules and priorities.

Impact

• •Improved equipment availability and reduced downtime
• •Enabled smoother onboarding and better collaboration through superior documentation
• •Increased overall group throughput (estimated ~40% from combined improvements)