Elite AI Academies &
HARNESS Pure-Rust Engine.
Master AI development across 13 interactive academies. Explore HARNESS (frontier pure-Rust inference engine running 30B–32B models on 8GB consumer VRAM) alongside 13+ free developer tools.
Master Gemini, OpenAI, Azure & Claude
Hands-on courses, labs, and quizzes built to match real search intent. Free with no sign-up, saved directly in your browser.
Zero to Hero Guide
Absolute beginner? Learn to set up VS Code, Python, and local AI models. Guaranteed to write your first simple app right away.
Vertex AI Academy
Free Google Cloud AI course covering Vertex AI, Gemini, context caching, grounding, and Agent Builder.
OpenAI Academy
Free interactive course to master ChatGPT, Custom GPTs, Responses API, Agents SDK, and reasoning models.
Azure Foundry Academy
Free Microsoft AI Foundry course covering model catalog, AI agents, RAG, safety, and enterprise deployment.
AWS Bedrock Academy
Free Amazon Bedrock course covering Runtime, Converse API, Guardrails, Knowledge Bases, Agents, and production AI operations.
Cursor Academy
Free Cursor AI course for Composer, Agent Mode, MCP, rules, and advanced AI coding workflows.
Claude Academy
Free Claude AI course with hands-on labs, prompt engineering, API, MCP, and 42 quizzes.
MCP Academy
Free Model Context Protocol course to build servers, clients, tools, and secure integrations.
Agents Academy
Free AI agents course for autonomous systems, multi-agent orchestration, tools, memory, and safety.
Open Source AI Academy
Free open-source AI course covering llama.cpp, vLLM, Ollama, quantization, LoRA, and MLOps.
Copilot Academy
Master Microsoft Copilot Studio, Cowork, M365 Copilot, Purview AI Shield & Power Platform.
Inference & Harness Academy
Free 18-module course covering custom Rust inference engines, GitHub Copilot runtime architecture, interference engineering, DeepSeek MLA & continuous batching.
Rust Systems & AI Academy
Free 20-module masterclass covering memory invariants, Cranelift JIT, bare-metal kernels, Void LLM & direct NVIDIA CUDA with cudarc.
Open Source on GitHub & Systems
Public repositories and foundational systems built from scratch: pure-Rust inference engines, anti-sycophancy gateways, self-hosting compilers, CUDA LLM trainers, bare-metal OS kernels, and MCP tooling.
Project SPINE
33-Vertebrae Anti-Sycophancy & Reality Gateway
Biological-cybernetic 33-vertebrae reality gateway in pure Rust. Eliminates LLM sycophancy, flattery traps, and unearned apologies (Thoracic Invariant T1) with <0.2 µs Aho-Corasick lexical scanning, two-phase streaming tail sanitization, and CoT ReasoningDemuxer.
ChronoFact
Epistemic AI Backbone & Token Optimization Engine
Native Rust 4-pillar engine for autonomous agents. Prunes 71.2% to 90.9% of prompt tool schemas via TF-IDF cosine similarity, intercepts weight freeze cutoff drift in <5ms, and gates semantic memory with a strict 0.72 threshold (0 leaked tokens). Native MCP server with live React 19 cockpit.
PulseBridge
Universal Mobile AI IDE Companion & Telemetry Bridge
High-precision, low-latency mobile companion for Google Antigravity, Cursor AI, and VS Code. Real-time desktop streaming via Windows GDI, touch-drag mouse wheel scrolling, live multi-project folder inspection (depth-2 tree), 7–30d recency filtering, zero-trust token persistence, and bidirectional prompt injection over 4G & Wi-Fi.
Infinity Signal
Open-Source 20-Indicator Trading Signal Dashboard
Next.js trading signal dashboard that computes 20 technical indicators from raw OHLCV bars via the Alpaca Market Data API, weights votes by market regime, and backtests setups in-browser. Read-only market data: it never places trades. Educational use only, not financial advice.
HARNESS Engine
Pure-Rust LLM Inference, wgpu GPU Kernels & Bio-SNN Runtime
Frontier LLM inference engine in pure Rust (11 crates, 15.7k LOC). Drives 30B Sparse MoE (Qwen3-Coder-30B-A3B) at 16.37–21.40 tok/s (14.11× speedup), 14B Speculative at 15.89 tok/s, and 32B Dense Speculative at 3.90 tok/s on an 8GB RTX 5060 with custom wgpu WGSL GPU kernels (948.6 GFLOP/s) and 24/24 compiler-verified pass rate.
Vitalis
AI-Native Self-Hosting Compiler
Compiled language built from scratch in Rust with Cranelift JIT + AOT, Hindley-Milner type inference, SIMD vectorisation, and self-hosting (compiles itself).
Void LLM
From-Scratch LLM Training & CUDA Engine
Custom transformer training and inference engine built from scratch in Rust with 11 hand-written CUDA kernels, 5 model configurations (5M to 3B parameters), and a 12-mutation self-evolution engine.
Freedom OS
Bare-Metal x86_64 OS Kernel from Scratch
Zero-dependency x86_64 bare-metal operating system kernel in Rust featuring a virtio-net driver, in-kernel TCP/IP stack, text-mode web browser, GUI window compositor, and NexusCode IDE.
The Forge
Multi-Agent Code Evolution Platform
Claude, GPT, Gemini and local models compete to solve coding challenges; the Vitalis JIT compiler judges by compiling and benchmarking the results, and evolutionary selection breeds the winners. Hardened for local use: key auth, path jailing, budget gates and fail-closed self-improvement.
MCPlex Gateway
Security-First MCP Smart Gateway
Open-source MCP smart gateway in Rust (v0.7.1) with security by design, constant-time auth, server-verified RBAC role binding, strict input validation, redacted audit logs, semantic tool routing (70–90% token savings) and an optionally-authenticated live security dashboard. Plugs into VS Code, Claude & Cursor.
AgentLens
Chrome DevTools for AI Agents
Time-travel debugging, cost tracking and anomaly detection for multi-agent LLM workflows. MCP-native, built in TypeScript.
Humaniser
Open-Source AI Text Humanizer (v1.2.0)
Zero-dependency vanilla ES-module AI text humanization & voice transformer engine. Strips zero-width watermarks & chat wrappers, rewrites mechanical n-gram clichés, injects sentence burstiness, and smooths perplexity across 6 voice modes.
Infinity Service Ops
Sovereign ServiceNow Alternative
Security-first, Rust-native service operations platform covering ITSM, CMDB/service graph, ITOM ingestion, SecOps, a workflow engine, and auditable AI. Own your critical-systems data.
Free Developer Tools: Instant Access
Project SPINE
33-Vertebrae Anti-Sycophancy & Reality Gateway
Pure-Rust biological-cybernetic reality gateway for frontier AI agents. Enforces 33 anatomical vertebrae (C1–Co4), Thoracic Invariant T1 against unearned apologies, and sub-microsecond Aho-Corasick streaming cushions stripper.
ChronoFact
Epistemic AI Backbone & Token Optimizer
Pure-Rust token optimization & epistemic backbone for AI agents. Slashes prompt tokens by 81.8% to 90.9% via TF-IDF schema pruning, intercepts weight freeze cutoff drift in <5ms, and gates memory with a strict 0.72 threshold (0 token leakage).
PulseBridge
AI IDE Monitor & Remote Companion
Monitor your AI IDE and coding agents from your phone. Zero-latency Win32 GDI screen streaming, touch-drag mouse wheel scrolling, live project file trees, 7–30d recency filtering & remote prompt injection for Google Antigravity, Cursor & VS Code.
Forge SEO
Meta Tag & SERP Snippet Generator
Generate perfect meta tags, preview Google SERP snippets & social cards with quality scoring.
ResumeForge
ATS-Ready Resume & CV Builder
Build professional resumes in minutes. AI bullet points, multiple templates & instant export.
ColorForge
Color Theory & WCAG Contrast Engine
Generate harmonious color palettes with 6 color theory modes. Contrast checker & CSS export.
SchemaForge
JSON-LD Rich Results Generator
Generate valid JSON-LD structured data for Google Rich Results. 12 schema types, one-click copy.
JSONForge
JSON Formatter, Validator & Diff
Format, validate, diff & explore JSON. Tree view, syntax highlighting, minify & beautify.
Prompt Studio
Agent Prompt & MCP Schema IDE
Build, test & optimize AI agent prompts with real-time scoring, MCP schemas, variable injection & 15+ templates.
ReadmeForge
Markdown & GitHub README Builder
Generate professional README.md files in seconds. 4 templates, shields.io badges & live preview.
CostForge
Multi-Model LLM API Cost Calculator
Compare Claude vs GPT vs Gemini API costs. 12 models, use-case presets & optimization tips.
Claude Toolkit
16 config files
Copy-paste config templates for Claude Code, Cursor, Copilot, Gemini CLI & Windsurf.
Humaniser
AI Text Humanizer & Detector Bypass
Convert robotic ChatGPT & Claude text into natural human writing. Injects burstiness, removes AI clichés & smooths perplexity.
Flagship Platforms & Commercial SaaS
Production commercial SaaS, EU AI Act regulatory intelligence, sovereign service operations, air-gapped enterprise appliances, and the Principal Architect engineering portfolio.
PropValuer Pro
Surveyors & Estate Agents PropTech SaaS
- ▸Instant UK property valuation running in <1ms with Landlord Delusion Index
- ▸Verified HM Land Registry comparable sales & EPC floor-area calibration
- ▸White-label RICS-style client PDF valuation dossiers generated in one click
EU AI Act Hub
UK & EU Regulatory Governance Platform
- ▸Interactive UK-vs-EU extraterritorial scope & high-risk classification checker
- ▸Article 113 enforcement timeline, GPAI systemic obligations & penalty calculator
- ▸Annex IV technical documentation & enterprise conformity assessment matrices
Infinity Security Stack
Pure-Rust Enterprise Cloud Infrastructure
- ▸Four Rust-native binaries replacing Auth0, Datadog, Pinecone, and Kafka
- ▸Infinity ID (OIDC/MFA), Observe (logs/traces), Data (HNSW), and Stream (BM25)
- ▸Auditable single-binary deployments with zero per-seat or per-GB SaaS tax
Infinity Service Ops
Sovereign ServiceNow Alternative
- ▸Security-first enterprise platform for ITSM, CMDB service graph & ITOM ingestion
- ▸Automated SecOps incident response playbooks with auditable AI agent workflows
- ▸Sub-millisecond Rust edge runtime with full on-premise data sovereignty
HARNESS Engine
Pure-Rust LLM Inference, wgpu GPU Kernels & Bio-SNN Runtime
- ▸Runs 30.5B Sparse MoE at 16.37–21.40 tok/s (14.11× speedup) & 32B Dense within 8GB consumer VRAM
- ▸Hardware-synced wgpu WGSL GPU kernels: 948.6 GFLOP/s GEMM, 313.9 GB/s Q4_K GEMV & 215 µs Sparse MoE
- ▸24/24 compiler-verified pass rate (rustc --test -O & Python DP) + DFA grammar-constrained decoding
About Bart & Projects
What I've Built Across Rust, CUDA & AI
- ▸Creator of SPINE, ChronoFact, HARNESS Rust Engine, Void CUDA MoE & Vitalis JIT
- ▸360K+ lines of from-scratch systems across Rust, CUDA, Compilers & Bare-Metal OS
- ▸13 free interactive AI academies & 13+ developer tools built for the community
Infinity Security Stack: Own Your Infrastructure
Production-focused alternatives to expensive SaaS in identity, observability, data/vector, and streaming/search.
Infinity ID
Auth0 / Okta replacement
Rust-native identity provider with OIDC/OAuth2, PKCE, TOTP MFA, RBAC, and rotating signing keys.
Infinity Observe
Datadog / Splunk replacement
Unified logs, metrics, and traces with full-text search, t-digest latency quantiles, and alert-ready telemetry.
Infinity Data
Snowflake / Pinecone replacement
Analytics plus vector database in one service with HNSW search, metadata filtering, and persisted indexes.
Infinity Stream
Kafka / Elastic replacement
Segmented commit logs, consumer groups, BM25 search, and real-time pub/sub in a self-hosted Rust runtime.
Explore full Infinity Security page
Migration playbook · comparison matrix · full technical breakdown
Questions, Answered
What is Infinity Tech Stack?
Infinity Tech Stack is an independent AI research and software portfolio by Bart Chmiel. It includes 13 free AI academies, 13+ free developer tools, and from-scratch systems research: Neuromantix AGI, the Vitalis compiler, the Void LLM engine, Freedom OS and the Infinity Security Stack.
Are the developer tools really free?
Yes. All developer tools (including the SEO meta tag generator, JSON formatter, schema generator, resume builder, colour palette generator, prompt playground, README generator, API cost calculator, Humaniser AI, and OpsForge Service Blueprint) are 100% free with no sign-up required.
Is the code open source?
Several projects are open source and published on GitHub, including Infinity Service Ops and the Infinity Security Stack (Apache-2.0), as well as Project SPINE, ChronoFact, PulseBridge, HARNESS, Humaniser, Infinity Signal, The Forge, MCPlex Gateway, and AgentLens (MIT). Each project page links directly to its public repository.
What is the Infinity Security Stack?
It is a suite of four open-source, self-hostable Rust services covering identity, observability, vector search, and real-time streaming, built as self-hosted alternatives to proprietary SaaS infrastructure. Infinity ID is stable; Infinity Observe, Data, and Stream are in active development.
Which technologies power the stack?
The underlying engines are written in Rust, with CUDA for GPU acceleration and TypeScript with Next.js powering the web platform and interactive tools.
Are the interactive AI academies free?
Yes. All 13 academies (covering Rust Systems & AI, Inference & Harness, Claude, OpenAI, Vertex AI, Azure AI Foundry, AWS Bedrock, Cursor, MCP, AI agents, open-source AI, Copilot & Power Platform, and Zero-to-Hero) are free interactive courses with no sign-up required, and your progress is saved locally in your browser.
What are the best free AI courses available right now?
Infinity Tech Stack offers 13 free AI academies in 2026: Rust Systems & AI, Inference & Harness, Claude, OpenAI, Vertex AI, Azure AI Foundry, AWS Bedrock, Cursor, MCP, AI Agents, Open Source AI, Copilot & Power Platform, and Zero-to-Hero. Each course includes hands-on labs, quizzes, and XP progression with no paywall or certificate fees.
How do I start learning AI coding as a complete beginner?
Start with the free Zero to Hero path at infinitytechstack.uk/zero-to-hero. It assumes zero prior knowledge, featuring an in-browser practice terminal, VS Code setup guides, line-by-line code walkthroughs, and a clear jargon decoder for technical terms.
Is there a free AWS Bedrock course?
Yes. The AWS Bedrock Academy is a free 25-module course covering boto3, least-privilege IAM, AgentCore, RAG debugging, guardrails, cost engineering, multi-region failover, and AI incident response, featuring real code, CLI labs, and scenario quizzes.
What is MCPlex?
MCPlex is an open-source MCP (Model Context Protocol) gateway published on GitHub under the MIT license. It routes and manages multiple MCP servers behind a single endpoint with semantic tool routing, RBAC and audit logging, and pairs with the free MCP Academy course for learning the protocol itself.
Are all courses and developer tools on Infinity Tech Stack completely free?
Yes. Every single one of the 13 AI academies (Rust Systems & AI, Inference & Harness, Claude, OpenAI, Bedrock, Vertex AI, Azure AI Foundry, Cursor, MCP, Agents, Open Source, Copilot & Power Platform, Zero-to-Hero) and all 13+ developer tools are 100% free with no sign-up, no subscriptions, and no paywalls.
What is Humaniser AI?
Humaniser AI is an open-source text humanizer and voice engine at infinitytechstack.uk/humaniser. It calibrates perplexity, burstiness, and natural tone to convert robotic AI writing into engaging human prose while bypassing mechanical AI detector patterns.
What is Project SPINE and how does it eliminate AI sycophancy?
Project SPINE (Structured Protocol for Invariant Neural Epistemics) is a pure-Rust biological-cybernetic 33-vertebrae reality gateway at infinitytechstack.uk/spine. It eliminates RLHF sycophancy, flattery traps, and unearned apologies (Thoracic Invariant T1) using sub-microsecond Aho-Corasick lexical scanning, a two-phase StreamTailSanitizer, and a CoT ReasoningDemuxer while preserving 100% of upstream KV prefix cache.
What is ChronoFact and how does it prevent agent epistemic drift and token waste?
ChronoFact is an open-source, native Rust epistemic AI backbone and token optimizer for autonomous agent workflows. It sits between agents and LLMs via Model Context Protocol (MCP) or an HTTP proxy. ChronoFact slashes prompt tokens by 81.8% to 90.9% through TF-IDF unigram schema pruning, intercepts knowledge cutoff and weight freeze drift in under 5ms, gates semantic memory with a strict 0.72 threshold (0 leaked tokens on task switch), and catches factual contradictions in 20ms using Natural Language Inference.
What is PulseBridge and how does it monitor AI coding agents?
PulseBridge is an open-source, ultra-low-latency remote companion and telemetry bridge for AI IDEs including Google Antigravity, Cursor AI, and VS Code. Built in native Rust (Axum 0.7) and React 19, it streams live desktop screens via Win32 GDI, provides touch-drag mouse wheel scrolling, live multi-project folder inspection (7–30 day recency filter), and bidirectional prompt injection over local Wi-Fi or secure Cloudflare tunnels directly from your phone.
Who is Bart Chmiel and what is Infinity AI?
Bart Chmiel is an AI research engineer and software architect based in London, UK. Infinity AI / Infinity Tech Stack is his independent research portfolio featuring from-scratch AGI research (Neuromantix), native compilers (Vitalis), LLM training engines (Void), bare-metal systems (Freedom OS), and comprehensive free AI education.