Portfolio

Real Tools.
Real Results.

Five integrated technologies built to save high-school coaches time and deliver verifiable insights from film and data.

The Simulator

Rep the look before Friday, and see the read get made

Twenty-two players running a real concept, with the quarterback's decision computed from the defender's actual movement rather than authored in advance. A run book of zone-read techniques and a pass book of eight concepts — drops, the route tree, ball flight and a progression the quarterback genuinely works — against seven defensive calls including slants, a T/E twist and a read-key stunt. Watch it as circles on a board, or in three dimensions down to hand placement and pad level.

Deterministic 30 Hz engineMeasured reads, never scripted
Explore →

Film Study

Automatic play clipping, resolution enhancement, annotation

Restoration, film diagnosis and field masking run today: the pipeline grades what you actually uploaded, cleans it, and isolates the playing surface. Automatic play detection, tracking and annotation are in active development. Every label a coach does get is reviewed before it counts, because a wrong label taught confidently is worse than none.

Film restoration (measured)Film diagnosis
Explore →

AI-ssistant Coach

Answers out of your concepts and your film, not out of memory

The assistant retrieves from the authored concept index and your labelled film before it answers, and it proposes changes rather than making them — a coach approves every one. It is self-hosted, so your film and your playbook never leave your control.

Retrieval over authored conceptsFilm-grounded answers
Explore →

Football Glossary

Every concept, authored by a coach and free to read

The index the whole portfolio is built on: what a concept is, why it matters, the coaching points, and what your program actually calls it. Written by coaches, never generated — which is what lets everything else in the portfolio cite it.

Authored indexProgram vocabulary
Explore →

HSSI — High School Sports Index

Numbers a college coach can trust, because film backs them

An athlete's profile with production verified against film rather than typed in by hand. Built for iOS and Android, and aimed at the player, not the program.

Kotlin MultiplatformiOS
Explore →

Every number here is traceable. Restoration is scored on a round trip where the truth is known — degrade a clean clip, restore it, measure the gap: SSIM 0.86 → 0.97, PSNR 21.2 → 33.4 dB. Helmet detection trains on a public-domain corpus of 9,947 frames and 193,736 boxes. Play detection, tracking and field homography are in development, and they are labelled that way wherever they appear.

No fabricated stats. No restricted marketing claims. Just technology coaches can trust today.