Disclaimer

In one sentence: everything on this site exists to demonstrate how AI systems are built — not to diagnose, advise, or replace a qualified professional in any field.

Educational purpose

RLcapstone.ai documents personal learning projects. The models, applications, and results described here were built to explore machine-learning techniques end to end. They have not been clinically validated, independently audited, or certified by any regulatory body, and they are not offered as products or services.

Not medical advice

Several projects (such as MoleCheck) involve health-related data. These projects are demonstrations of image and signal classification, nothing more. They are not medical devices, they cannot diagnose any condition, and their outputs must never be used to make a health decision.

If you have any concern about a mole, a skin change, or any other health matter, consult a physician or the appropriate specialist. A model's "benign" output is not reassurance, and a "flagged" output is not a diagnosis — in either direction, only a qualified clinician can determine what a finding actually represents.

Not financial advice

The stock-volatility project is an educational machine-learning exercise. It predicts volatility, not price direction, and nothing on this site constitutes a recommendation to buy, sell, or hold any security. No one involved is a licensed financial adviser.

Accuracy and limitations

Reported metrics (accuracy, sensitivity, ROC-AUC, and similar) describe performance on specific public research datasets under specific conditions. Real-world performance — with different cameras, lighting, skin tones, patients, or market regimes — can be substantially worse. Machine-learning models fail in unpredictable ways, and these write-ups document those failure modes deliberately, because understanding them is central to the educational purpose.

Security content

The penetration-testing project and its defensive extension are strictly educational and operate only within an isolated, self-contained sandbox on the author's own machine, using standard OWASP Top 10 teaching examples. Such techniques must only ever be applied to systems you own or are explicitly authorized to test.

No warranty

All content, code, and models are provided "as is", without warranty of any kind. Use of anything published here is at your own risk. The author accepts no liability for decisions made, or actions taken, based on this site's content.

Data and privacy

These projects favor on-device inference: where a demonstration or application processes an image or signal, the processing occurs locally and the data is not uploaded to any server. Datasets used for training are public research datasets (for example, the ISIC Archive and MIT-BIH), used under their respective licenses.