Tony W.

I eliminate $10k-50k/month in cloud API costs by deploying private AI infrastructure that runs on your hardware without sending data to third parties. Recent work: Built an invoice parser for an Australian financial auditor processing 62 documents with 100% accuracy on totals and 92.5% on vendor extraction. System runs entirely on-premise with zero external API calls. Team got working endpoints in 4 days. What I deliver: - Invoice/document automation (OCR + LLM extraction) - On-premise LLM deployment (vLLM, Ollama, local models) - Government contract automation (SAM.gov, DIBBS integrations) - SOC2-compliant RAG systems (ChromaDB, Qdrant, pgvector) I own dual Tesla P40 GPUs (48GB VRAM) for rapid prototyping and testing. Every system I build is production-grade - I handle Linux optimization, Docker orchestration, and deployment from silicon to API.