The sixth entry in a live builder's log. The contribution I was certain would be the centerpiece lost to the cruder approach already in the pipeline. I published the loss, because complexity has to earn its place every time.
A deeper Builder Journal entry from inside the Hyperspectral Object Tracking Challenge 2026, with code. Before a single model could matter I had to build a local scorer for an unlabeled test set and beat a stale submission template. The plumbing is half the competition.
The fifth entry in a live builder's log. A parameter sweep handed me a new best score. Three red flags said it was noise wearing a knob's clothing, so I refused to ship it. Most of competitive ML is declining to fool yourself.
The fourth entry in a live builder's log. I built a faithful local scorer, then fed it a biased test set, and it confidently signed off on a change that made me worse on the real leaderboard. A dev set built from failure cases certifies rescues, not deployments.
The third entry in a live builder's log. My first Kaggle submission was rejected before it ever scored, thanks to a stale sample_submission file and a Google Drive download limit. The plumbing turns out to be half the competition.
The second entry in a live builder's log. The Hyperspectral Object Tracking Challenge scores you on an unlabeled test set, so the leaderboard can't tell you your own score. Here's how I built a local scorer byte-faithful to the official one, and why measurement is half the battle.
The first entry in a live builder's log. I'm competing in the Hyperspectral Object Tracking Challenge 2026: track one object through video shot in colors the human eye was never built to see. Here's what the problem is, why it's hard, and where I'm starting on the leaderboard.
The results piece has the metrics table. This is the other half: the decisions, the forks I parked, the harness I built before training anything, and the negative result I published so nobody re-runs it. Director's commentary on building a solo ML pipeline.
In March 2025 a lander reached the lunar south pole and tipped over. That gap between the most valuable ground in the solar system and the hardest to read is the thing I couldn't put down, so I spent a few months teaching a single laptop to scout it.
ALAN SCOTT ENCINAS CASE STUDY · LUNARSITE (FULL PIPELINE) LunarSite: An end-to-end ML pipeline for lunar south pole landing site selection Sim-to-real terrain segmentation, fine-tuned crater detection, PSR-aware XGBoost site scoring, and calibrated uncertainty — validated against NASA’s Artemis III…
How a flower classification model became a medical diagnostic tool for underserved communities. Using transfer learning with InceptionV3, we built a pneumonia detection system that achieved 73% validation accuracy and increased screening throughput by 340% at UCI Medical's community health center. Real deployment, real impact, real constraints.
