Projects · 16 systems
Systems that operate themselves.
Data and AI systems I designed and built. Filter by stack, pattern, or domain. Every entry links to a case study with an architecture diagram and says plainly whether it is in production, a pilot, or a prototype.
Async ETL Platform● In productionWorkforce Report-Compliance Platform● In productionQuote Automation● In productionPipeline Guardian● In productionDARA: Data Analyst Reporting AgentPrototype, in developmentEvent-Driven Sync on Google Cloud● One job in production · listener in shadowReport Automation● In productionCross-Source Date Validator● In productionGmail Document Parser● In productionReport Run Ledger and SchedulerPilot (test phase)RFDS Extractor● In productionPDF Attachment Extractor● In productionGC Asset LakePrototypeDaily Claude DigestPersonal toolOps PortalPrototypeAgent TownPrototype
Async multi-pipeline ETL into Postgres: about 14 scheduled pipelines across 25 GitHub Actions workflows
Internal platform for daily-report approvals, reminders, management PDFs and extracts, with write-back to the vendor API as each approver
Internal quoting tool that took accounting from 20-30 quotes a day to over 100: review queue, batch PDF-to-Drive and Gmail-draft emailing
Email-conversational ETL remediation agent with human-in-the-loop approvals
Chat-first NL→SQL analytics with Postgres RLS and a defined-metric library (prototype)
Change events over SSE trigger debounced incremental walks and guarded idempotent upserts, with an hourly reconcile as the safety net
Scheduled report suite: finance, compliance and open-items reports via Gmail, Drive and Google Chat
Daily reconciliation of task completion dates in the system of record against dates reported by email
HTML email → JSONB with dynamic field discovery
Pilot: one place that records every report run, dispatches scheduled ones, and flags missed or stuck runs
Gmail PDF scraper with a deterministic parser + optional Claude fallback
Org-wide parallel PDF requirement downloader over the vendor's project-management API
Lakehouse-lite ETL: incremental API sync with hybrid hot (Postgres) / cold (Parquet + DuckDB) storage
Personal daily briefing: fail-soft fetchers, one Claude composition call, sent over Gmail, scheduled via GitHub Actions
Internal ops portal UI: unified data-health, dashboards & embedded analyst (foundation phase)
Live pixel-art town that visualizes a monorepo's activity in real time