What it proves
Proof · Case Studies
What the work
What the work
proves.
Selected case studies showing how I approach real workflow problems: identifying the bottleneck, designing a practical solution, validating the result, and making it usable for the people involved.
Selected case studies.
01
Production Media Library Browser & Indexer
RootFX: Production Media Library Browser
Problem
Media libraries and production folders become difficult to search, preview, and manage safely as they grow.
Solution
A controlled browser and indexer with a FastAPI backend, SQLite + Meilisearch index, React UI, safety restrictions, and tool-managed cache/index folders.
Safety design
Read-only library roots, resolved-path allowlist checks, controlled API access instead of raw directory serving, and clear separation between index and source.
Stack
FastAPI + SQLite + Meilisearch + React, with a Nuke/plugin bridge direction for in-tool access.
Result
Designed for VFX/media libraries — safer search and preview over large asset roots without exposing the filesystem.
What it proves
Pipeline-safety thinking, API design discipline, search-system understanding, and tooling that respects production permissions.
02
Server-to-Server Copy Controller
S2SC: Safe Server-to-Server Copy Controller
Problem
Large production transfers routed through workstations are slow, risky, and hard to verify.
Solution
A server-to-server transfer controller using rsync over SSH with dry-run review, allowlisted paths, conflict handling, logs, and optional verification.
Safety design
Source server sends directly to destination, dry-run prep before copy, no deletes, no overwrite by default, optional checksum verification, and per-job logs and monitoring.
Stack
Python orchestration, rsync over SSH transport, FastAPI control surface, structured logs.
Result
Safer data movement, less manual handling, and a clearer audit trail for production transfers.
What it proves
Data-workflow thinking, infrastructure understanding, production-pressure handling, and an automation mindset grounded in safety.
03
Color Pipeline Diagnostic Tool
ChromaScope: Color Pipeline Diagnostic Tool
Problem
Color pipelines are hard to understand because gamut, transfer functions, white points, display transforms, and previews are often mixed together.
Solution
A technical diagnostic tool that separates color concepts and visualizes them clearly for learning, analysis, and debugging.
Diagnostics
Gamut geometry, transfer curves, white-point adaptation, diagnostic previews, with exact math separated from preview visualization.
Stack
React + TypeScript, color-science logic, CIE diagrams, ACES/OCIO thinking, with a Web/Tauri direction for desktop use.
Result
A practical educational and diagnostic tool for media-pipeline analysis and color-workflow learning.
What it proves
Color-science literacy, ability to teach complex systems, and front-end engineering applied to a hard technical domain.
04
AI R&D: Pipeline Risk Analysis
Pipeline Intelligence Lab: AI R&D for Render Risk
Problem
Render farm failures, slow frames, IO bottlenecks, worker anomalies, and delivery issues are often discovered too late.
Solution
Research prototype for turning logs into structured risk signals and recommendations across a cause library.
Cause library
Memory risk, IO bottleneck, worker anomaly, first-frame penalty, color transcode risk, and review generation risk.
Status
Active R&D — not production-ready. No accuracy or production-result claims. Real production logs stay private; synthetic data is used only for public demos and testing.
Stack
Python, ML R&D tooling, synthetic log datasets, structured signal extraction.
What it proves
Research mindset, willingness to label work honestly, and a structured way of thinking about pipeline failure patterns.