I build production trading systems, LLM platforms, and quantitative ML for hedge funds, banks, and insurers. Not a pitch deck - shipped infrastructure, inside institutions you've heard of.
Every engagement delivers production systems - not proofs of concept. Scoped for measurable results within 90 days.
GPU-accelerated deep reinforcement learning trading systems. I've built 144+ RL-based expert advisors spanning transformers, state-space models, spiking neural networks, and diffusion models - covering scalping, swing, and position strategies on equities, forex, and commodities.
Enterprise inference platforms on Kubernetes with zero-trust security, GitOps deployment, and full observability. Retrieval-augmented generation for financial documents - compliance, risk, regulatory filings - with source-grounded accuracy your team can trust.
Deep-dive security and architecture reviews of existing AI systems. I've caught 87 critical deficiencies in a single enterprise platform and rebuilt it from scratch - microservices, GPU-accelerated OCR, async processing, full audit trails for regulatory compliance.
Representative engagements across banks, trading, and insurance. Details sanitized for confidentiality.
Built a comprehensive library of GPU-accelerated reinforcement learning expert advisors on the NeuroNet_DNG framework (109K+ lines, 334 classes, 250+ neuron types). Architectures span transformers, Mamba SSMs, spiking neural networks, diffusion models, and graph networks - deployed for scalping, swing, and position trading across forex, commodities, and indices.
Architected GPU-accelerated inference infrastructure on Kubernetes for a major financial institution - 4×NVIDIA L40S GPUs, dual inference engines (SGLang + vLLM), GPU time-slicing, GitOps via ArgoCD, zero-trust security with HashiCorp Vault, and full Prometheus/Grafana observability.
Audited an enterprise document management system, uncovered 87 critical deficiencies, then led a ground-up rebuild with NVIDIA H200 GPU-accelerated OCR, async task queues, RBAC, and full audit logging - critical for insurance claims processing and regulatory compliance.
Designed and built JurisGraph - a forensic intelligence platform on Neo4j, LangChain/LangGraph, and FastAPI for mapping corporate relationships, financial flows, and regulatory exposure. Integrated Supreme Court of Canada case law with fine-tuned LLMs using QLoRA SFT and activation steering.
Tell me what you're building. I'll respond within 48 hours with an honest take on whether I can help - and if I can't, I'll point you to someone who can.