ValerioDolci

Valerio Dolci

Quantitative risk modeler & physicist who builds.
PhD in physics → A-IRB credit-risk models (LGD/EAD) in banking. Off-hours I
design small, complete tools and ship them — most orchestrated end-to-end with
AI agents.

I care about two things: models that hold up under scrutiny, and software that
actually works in someone's hands.

---

Selected work

| Project | What it is | Stack |
|---|---|---|
| slidewright ⭐ | Visual editor for HTML decks — drag&drop reorder, WYSIWYG text & shape editing, HTML/PDF 16:9 export. 100% local, single file. | Vanilla JS · Vite |
| snapsolid | Photogrammetry pipeline: photos → printable STL. | Python |
| hex-tactics | Turn-based tactics game with a reinforcement-learning AI opponent (self-play, MAPPO). | TypeScript · Python |
| tiny-reasoner | Experiments on uncertainty & reasoning in small neural nets. | Python |

---

Background

• Credit-risk modeling — internal technical referent for LGD & EAD A-IRB

model streams: development, recalibration, validation support and model governance. Practical AI adoption for risk management.

• PhD, Accelerator Physics (Sapienza Rome) — THz generation, ultrafast

lasers, non-linear optics. Peer-reviewed publications.

• Currently exploring AI governance & the EU AI Act applied to banking.

Toolbox: Python · SAS · SQL · MATLAB · C/C++ · TypeScript · Bash — plus agentic workflows as a daily multiplier.

---

*Biella, Italy · physics, credit risk, and things that compile.*
*Reach me: [email protected]*

Collapse