Full-stack systems builder focused on backend-heavy products, automation pipelines, AI-assisted workflows, and security-minded debugging.
Portfolio | Projects | npm CLI PR
I like useful systems with real moving parts: APIs, workers, queues, data models, logs, documentation, and the small operational details that make a project understandable after the first weird failure.
Most of my work sits around:
- backend-heavy products and APIs
- AI-assisted workflows with tools and memory
- automation pipelines for repeated work
- dashboards and developer-friendly interfaces
- debugging, evidence, and reproducible technical notes
Frontend: React, Vite, Next.js, Tailwind CSS
Backend: Django, Django REST Framework, Node.js, Express, FastAPI
Data and infra: PostgreSQL, Redis, Prisma, Docker, Celery, WebSockets
AI and automation: LLM workflows, agents, GitHub Actions, Discord pipelines, vector search
Debugging: logs, health checks, source review, proof notes, root-cause analysis
| Project | What it does | Stack |
|---|---|---|
| FinWise | Finance workspace for budgets, goals, charts, and decision support. | React, Django, REST APIs, data visualization |
| Lumina | Wellness and mental-health oriented platform with a clean product surface and practical user flows. | React, backend services, product UI |
| Real-Time Price Tracker | Monitoring workflow for price changes, alerts, and repeated data checks. | Automation, scraping, tracking logic |
| Django REST Boilerplate | Reusable Django REST starting point for authenticated APIs and structured backend development. | Django, DRF, API patterns |
| Team Task Manager Demo | Small collaborative work-management demo focused on CRUD flows and usable UI states. | React, API workflows |
Some work is more experimental and less product-shaped. That is where I explore agent behavior, automation design, and security research workflows.
- PA-Agent: Discord-to-GitHub-Actions-to-LLM knowledge pipeline for collecting, processing, and reusing structured context.
- Multi-Agent Deliberation: experiment around blind-first vulnerability assessment, independent reasoning, and consensus-style review.
- Reporter / TF-Reporter: research memory and reporting workflow for tracking security findings, notes, evidence, and follow-up context.
- npm CLI PR #9473: contributed a registry path validation fix in npm Arborist's allow-remote enforcement path.
I prefer building in thin, testable layers: first make the behavior clear, then make it reliable, then make the interface feel calm enough to use repeatedly. I care about naming, notes, traceability, and leaving enough context that future debugging does not start from zero.
Right now I am especially interested in backend systems, AI agents, automation infrastructure, reliability debugging, and developer tools.