
CountBubble — Nonprofit Case Management & Custom Reporting Engine
Maintaining a live SaaS platform, refactoring legacy architecture, and engineering a persistent multi-stage reporting workspace.
Experienced in taking products from a blank repo to stable deployment, setting up automated scraping pipelines, and building AI systems that go well beyond basic API prompts.
Real-world applications built under live constraints, ranging from full-stack SaaS and async AI engines to large-scale data systems.

Maintaining a live SaaS platform, refactoring legacy architecture, and engineering a persistent multi-stage reporting workspace.

Building a full-stack book generation system from scratch with Google Sheets outline parsing, asynchronous translation pipelines, and verified Google Cloud OAuth.

Building an API-controlled data collection service with dual-mode HTTP and browser automation pipelines, dynamic scheduling, and automated proxy monitoring.

Extracting customer reviews across a massive hotel queue with Playwright API interception, proxy bandwidth optimization, and a client management control panel.

Implementing delivery class management, coupon validation engines, and checkout state stabilization on a live multi-vendor platform.
Frameworks, databases, and AI tooling I rely on to take applications from zero to stable production.
Building autonomous agent loops, deterministic workflows with structured outputs, and RAG pipelines.
Developing clean REST APIs, server-side business logic, and type-safe user interfaces designed for production.
Designing relational schemas, document stores, async queues, and containerized server deployments.
Engineering resilient dynamic extractors, proxy-rotated crawlers, API interceptors, and cron workers.
Side projects, open-source utilities, and software tools built to solve specific workflow frictions.
A voice-first workspace that transcribes audio, auto-generates structured summaries and tags, and indexes notes for an interactive RAG assistant using dense vector similarity (pgvector) and sparse keyword search fused with RRF.
A distraction-free Markdown workspace with in-place AI writing assistance, visual inline diff reviews, and command-palette search. Features bidirectional Markdown sync, voice transcription, zero-config SQLite/PostgreSQL persistence, and offline demo simulation.
Reflections on software development, open-source culture, learning in public, and building reliable systems.
I’m currently available for freelance engineering work, technical consulting, and full-stack or applied AI projects. Send a message and I’ll get back to you promptly.