AI Enablement Specialist · Imobisoft
Sole builder of the branch's internal tooling — shipped full-stack products solo (concept → live) with AI-assisted and agentic workflows, balancing rapid delivery with security and reliability.
Highlights
- Agentic Workflow R&D: Used BulkUp — a cross-platform mobile app (Expo/React Native + TypeScript + Supabase; 6 screens, 5 migrations, offline-first sync, zero free-tier cost via EAS Build) — as the field test for an orchestrated multi-agent coding workflow, producing measured findings (~4x delegated-vs-single-session token cost, reviewer-agent as highest-ROI gate, orchestration gaps) written up into a reusable methodology (unified & orchestrated workflow docs, ready-to-copy subagent templates) now structuring the branch's engineering delivery.
- AI Integration: Engineered a photo-to-macro scanner with a server-side LLM proxy (Deno Edge Function + Llama-4 vision), JWT gating, structured output, and safe handling of untrusted model outputs.
- Security & Compliance: Enforced Row-Level Security with column-level grants, validated by pgTAP (44/44 tests passing) and live Playwright E2E tests against the production database.
- Agentic Systems: Built an orchestrator with 5 custom subagents; created TheBench (Claude Code upskilling plugin) and TheCompanion (Python cost tracker, 66 tests) adopted across a 16-engineer, 5-team org via direct team interviews.
- Design & UX: Designed interactive scoring UIs and a schema-driven frontend ranking 87 models (Vite + React + Tailwind v4 + shadcn/ui); authored decision records and prompt playbooks for team scaling.
Technologies
- Expo / React Native
- TypeScript
- Supabase
- Llama-4 Vision
- Playwright
- Claude Code