∞
I'm a London-based engineer focused on low-level AI infrastructure, compilers, and native Rust systems. Over the past few years I've built Project SPINE (33-vertebrae anti-sycophancy gateway), ChronoFact (epistemic AI backbone), HARNESS (pure-Rust 30B–32B inference), Vitalis v1333 (self-hosting Cranelift JIT compiler), Void LLM (CUDA MoE trainer), and Freedom OS (bare-metal x86_64 kernel).
I learn best by building the whole stack myself with no black boxes. Across 360,000+ lines of code in native Rust, custom CUDA kernels, Cranelift JIT compilers, and web tooling, every project here started as a hands-on deep dive into how systems actually work at hardware speed. Alongside my core research, I built 13 free interactive AI academies and 13+ browser-native developer tools on Infinity Tech Stack so anyone can learn and ship with zero paywalls or sign-ups.
Six core systems written in native Rust, CUDA, Cranelift JIT, and TypeScript, complete with open-source repositories, automated test suites, and live deployments.
33-Vertebrae Anti-Sycophancy & Epistemic AI Backbone
Security-First Rust MCP Router & Zero-Trust Control Plane
Chrome DevTools for AI Agents & Multi-Agent JIT Evolution
Pure-Rust 30B–32B Inference Runtime & Custom CUDA LLM Engine
Self-Hosting Cranelift JIT Compiler & Bare-Metal x86_64 Kernel
Mobile AI IDE Telemetry Bridge, Sovereign ITSM & Free Education
Compilers, kernels, inference runtimes, security gateways, and free learning platforms built by hand.
33-vertebrae biological-cybernetic epistemic firewall, multi-tier reality dial, and pushback verification in pure Rust for autonomous AI agents. Live React 19 telemetry HUD.
Native Rust temporal cutoff calibration, AST code security & NLI claim verifier, zero-pollution memory, and TF-IDF cost optimizer for frontier AI models.
Frontier Pure-Rust LLM Inference Engine & Native VS Code Agent. 30B Sparse MoE at 16.37–21.40 tok/s & 32B models on 8GB VRAM, wgpu WGSL GPU kernels, LIF Spiking Attention.
From-scratch LLM training & inference engine written in pure Rust + CUDA. Features a 12-expert Mixture of Experts architecture, Top-2/Top-4 router, 32K BPE tokenizer, and 9,500 tok/s GPU throughput.
Production IT operations management & enterprise service observability platform featuring SOE/AOE service blueprints, incident telemetry, and real-time infrastructure health tracking.
Interactive UK & EU AI Act regulatory compliance checker & legal scope evaluator. Calculates risk classification, enforcement timelines, and technical requirements.
Self-hosting compiled programming language built from scratch in Rust. Features Cranelift JIT/AOT code generation, Hindley-Milner static type inference, and AST-level program analysis.
13 free interactive learning platforms: Claude, OpenAI, Cursor, Bedrock, Vertex AI, Azure Foundry, Agents, Open Source AI, Copilot, Rust Systems, Harness, MCP, and Zero to Hero.
13+ free production developer tools: Project SPINE, ChronoFact, PulseBridge, Humaniser AI, Forge SEO, ResumeForge, ColorForge, SchemaForge, JSONForge, CostForge, Prompt Studio.
Security-first MCP smart gateway in Rust. Semantic tool routing cuts context overhead by 70–90%, with constant-time auth, server-verified RBAC, and a live security dashboard.
Chrome DevTools for AI agents and multi-model evolution arena. Real-time WebSocket tracing of tool calls, token usage, decision trees, and JIT-verified pipelines.
Custom x86_64 operating system kernel written in bare-metal Rust. Implements physical memory paging, preemptive multitasking, APIC timer interrupts, and a zero-dependency TCP/IP networking stack.
30+ languages, compilers, and infrastructure tools used across my projects.
Browse the source code on GitHub, connect with me on LinkedIn, or send me an email to chat about Rust, AI systems, or engineering.