Evan Freed
AI and Machine Learning Projects

Evan Freed — AI and Machine Learning Projects

A focused collection of end‑to‑end AI/ML prototypes—each one designed to be easy to evaluate quickly in an interview setting: clean UI, real functionality, and clear tradeoffs. The emphasis here is practical engineering: shipping demos, wiring backends, and turning “AI ideas” into tools that can be used, tested, and iterated.

The three flagship projects below cover: (1) multi‑LLM orchestration and structured debate, (2) prompt engineering workflows and coaching loops, and (3) classic ML inference in the browser (handwritten digit recognition).

Featured Projects

Protagoras

Multi‑LLM debate orchestrator with synthesis + similarity matrix.

Prompt Builder v2.17

Prompt engineering workbench: build → test → coach critique → iterate.

Using Machine Learning for Recognition of Handwritten Digits

MNIST Digit Classifier — Draw → Predict: draw a digit (0–9) and run in‑browser inference.

Latest posts

Short, practical write-ups: what I built, how it works, what I’d do next.

2026-01-15Case study

Protagoras v1.5.0: Multi‑LLM Debate + Similarity Matrix

A deployable debate orchestrator with multi-provider participants and an optics-first similarity matrix.

2026-01-15Build note

Handwritten Digit Recognition: Draw → Predict in the Browser

How the MNIST demo works end‑to‑end: canvas input, preprocessing, and in‑browser model inference.

Coming soonEngineering note

Prompt Builder: from template to testable prompt

A practical workflow for building prompts you can evaluate, refine, and reuse—plus what “Prompt Coach” feedback looks like.