Aaryaman Shinde

I'm completing a Mechanical Engineering degree while building AI systems that are measurable, testable, and reliable. My work spans Generative AI evaluation, automated validation, data pipelines, and predictive maintenance.

At Tata Power, I processed 2M+ solar data points, built a Streamlit SCB monitoring dashboard with anomaly detection, and created a degradation analysis tool that enabled 30% more accurate maintenance scheduling. I also automated data pulling, cleaning, and validation with Python and tested dashboard behaviour using synthetic edge cases.

In AeroTwin, I built a physics-informed digital twin for jet-engine RUL prediction that reduced RMSE by 57% against a Random Forest baseline. For AutoCNC, I designed a multi-agent CNC workflow generator and a reproducible six-metric evaluation framework, using 75 test cases and blind dual-expert review to quantify model reliability.