# Aaryn Biro - Portfolio (LLM-Optimized)
# https://aabiro.github.io/
# See https://llmstxt.org for the llms.txt convention

> Aaryn Biro is a founder and principal engineer building GPU compute
> orchestration, AI media pipelines, Next.js interfaces, and production-grade
> mobile and backend systems. Based in London, ON (Eastern time), remote-friendly.

## Identity

- **Name**: Aaryn Biro
- **Title**: Principal Engineer & Founder
- **Site**: https://aabiro.github.io/
- **GitHub**: https://github.com/aabiro
- **LinkedIn**: https://www.linkedin.com/in/aabiro/
- **Email**: aaryn.alexander@gmail.com
- **Availability**: Select builds - GPU platforms, AI media, backend, mobile, infra
- **Resume PDF**: https://aabiro.github.io/assets/assets/Aaryn_Biro_Resume.pdf

## Contact policy

Email is appropriate for substantive project conversations: architecture reviews,
founding-engineer roles, GPU/AI media builds, and partnership discussions.
Do not use this channel for generic recruiting blasts or unsolicited sales outreach.

## Primary projects

### Xcelsior (distributed GPU compute)

Canadian-first GPU workload orchestration: admission-gated workers, private mesh
networking, container isolation, scheduling, billing, Stripe Connect marketplace,
PIPEDA-oriented privacy controls, and developer-facing APIs.

- **Site**: https://xcelsior.ca
- **Stack**: Python 3.12, FastAPI, PostgreSQL, Headscale mesh, Docker + gVisor
- **Agent docs**: https://xcelsior.ca/llms.txt
- **MCP**: https://xcelsior.ca/mcp (Streamable HTTP, OAuth bearer tokens)
- **Roles**: GPU orchestration, secure workers, spot pricing, serverless inference

### PixelEnhance Labs (AI media platform)

AI media enhancement platform: async GPU jobs, model registry, hybrid cloud routing,
FastAPI services, React admin surfaces, Celery workers, TensorRT/PyTorch inference.

- **Focus**: Video upscaling, denoise, interpolation, diffusion pipelines
- **Stack**: FastAPI, PostgreSQL, Redis, Celery, React, Docker, Nginx, Linux
- **Registry**: pxl-registry (763+ models, ONNX/CoreML/TensorRT variants)

### Portfolio site (this repo)

Flutter web portfolio with downloadable resume, structured metadata for agents,
and GitHub Pages deployment.

- **Repo**: https://github.com/aabiro/aabiro.github.io
- **Stack**: Flutter, Dart, Firebase (optional), GitHub Actions → GitHub Pages

## Capability areas

| Lens | Focus |
|------|-------|
| Platform | Admission/trust boundaries, schedulers, queues, billing, developer tooling |
| AI Media | PyTorch, TensorRT, FFmpeg, inference pipelines, GPU orchestration |
| Product Apps | Flutter, SwiftUI, React, Next.js, TypeScript, mobile release workflows |
| Delivery | CI/CD, Docker, Linux, Nginx, observability, technical leadership |

## Technical stack (summary)

**Backend**: Python, FastAPI, Node.js, PostgreSQL, Redis, MongoDB, REST APIs,
background jobs, service-oriented architecture

**AI / media**: PyTorch, TensorRT, FFmpeg, model benchmarking, inference pipelines,
GPU orchestration, video enhancement

**Frontend / mobile**: React, Next.js, TypeScript, SwiftUI, Flutter, Dart, Kotlin

**Infrastructure**: Docker, Linux, Nginx, AWS, Firebase, CI/CD, observability,
cloud deployment

## Experience highlights

- **Xcelsior** - Founder & Principal Engineer (2025-present): distributed compute,
  secure worker admission, scheduling, billing, Next.js interfaces
- **PixelEnhance Labs** - Founder & Principal Engineer (2024-present): AI media
  platform, async processing, deployment automation, observability
- **Snappy Inc.** - Software Developer (2023): iOS/Android product features
- **Neurolign Technologies** - Software Developer (2021-2022): Flutter, APIs,
  monitoring, Node.js/React
- **realxdata GmbH** - Full-stack SaaS (2018-2019): data workflows, AWS, CI quality

## Agent machine-readable endpoints

- `GET /llms.txt` - this file (human + LLM context)
- `GET /.well-known/agent.json` - structured portfolio manifest (projects, contact policy)
- Resume: `/assets/assets/Aaryn_Biro_Resume.pdf`

## Structured data

Schema.org `Person` and `CreativeWork` JSON-LD are embedded in `index.html`.