Preballin / @preballin
I'm Thaddeus Arndt, an AI workflow builder, creative automation operator, and music/audio technologist in Los Angeles. This page collects public-safe project work, GitHub demos, and case-study directions without exposing private credentials, account data, raw logs, or production-only systems.
Three clean-room GitHub demos showing tested, public-safe systems for research automation, music/audio operations, and approval-gated agent workflows.
Public-data research workflows for content gaps, market monitoring, narrative scanning, opportunity discovery, redaction, approval queues, and structured reporting.
The public repo intentionally uses synthetic fixtures and report-only logic. It excludes private strategy, credentials, account details, wallets, live execution, and production-only configuration.
View on GitHubMusic/audio workflow utilities for synthetic track metadata, release planning, prompt-safe content repurposing, markdown reports, tests, and CI.
The public repo uses synthetic metadata only. It excludes real audio assets, stems, masters, paid loops, private client data, platform credentials, and unsupported placement or performance claims.
View on GitHubSimulated clean-room workspace patterns for mock task routing, tool registries, dry-run orchestration, policy gates, blocked actions, and reviewable trace reports.
The public repo avoids private assistant infrastructure. It excludes real hostnames, raw chats, assistant memory, platform sessions, production jobs, and live external actions.
View on GitHubAll three demos are structured around safety boundaries: synthetic examples, no private logs, no secrets, no live account mutation, no sensitive source lists, no production credentials, and no unsupported claims.
These are sanitized directions being prepared from real work, with private implementation details removed before publication.
Case-study patterns for systems that research, draft, prepare, and report while stopping before sensitive submissions, payments, account changes, or external messages.
Static-site funnels, booking pages, portfolio systems, SEO content workflows, and creator commerce patterns.
Public-safe app concepts for theater/scene practice, rehearsal flows, mock script data, and mobile-ready product notes.
Read-only dashboard and reporting patterns using synthetic data, with no wallet code, trading execution, private thresholds, or account-specific details.
Some of my strongest work involves private workflows, client/account contexts, and automation systems that should not be copied directly to GitHub. My public portfolio approach is to extract the architecture, tests, documentation, synthetic examples, and safety patterns while leaving private edge, credentials, logs, and production systems out of public repos.