01 / 6 PROJECTS

Enterprise AI Platforms

Platform infrastructure for AI at organizational scale: retrieval, gateways, and agent tooling, built and operated in production.

01 Enterprise AI Platforms

AIR API — AI Retrieval API

An AI-driven retrieval platform that serves grounded answers from corporate knowledge sources (Help Centre + cTrader Admin Guide) to any AI assistant, support automation, or team workflow inside Spotware — through a REST API and a standards-compliant MCP server, with automatic twice-daily content refresh.

Active — Present
02 Enterprise AI Platforms

YouTrack MCP Server

A production MCP server built at Spotware that gives AI assistants deep, agent-optimized, multi-account access to YouTrack — issues, Knowledge Base articles, project schema, comments, and commands, read and write. It makes oversized specifications, tickets, and discussion threads navigable instead of unreadable, turns a failed call into a correction that teaches the fix, and adds batch fetch, change history, attachment access, and ChatGPT Deep Research / Company Knowledge compatibility.

Active — Present
03 Enterprise AI Platforms

Claude Code Gateway

The corporate LLM gateway between every engineer's coding agent and the upstream model providers: provider routing and failover across multiple AI labs, per-user request, token, and cost accounting at API prices, an authenticated admin surface with usage analytics, and centrally administered usage governance — rebuilt from an open-source foundation into a corporate-grade service.

Active — Present
04 Enterprise AI Platforms

MCP Gateway

The corporate MCP gateway on ToolHive: one authenticated, policy-governed entrance where AI clients reach internal tool servers — corporate sign-in through OAuth 2.1, per-user authorization evaluated on every tool call that passes through it, audited invocations, and the defects found along the way fixed upstream in the open, with a first merged pull request credited in the ToolHive v0.46.0 release notes.

Active — Present
05 Enterprise AI Platforms

Managed Agent Platform

A self-hosted platform for running AI agents in production rather than as one-off scripts, in active development as one product: a configurable Deep Agents runtime where a new assistant is a stored configuration, an isolated code sandbox with credential injection kept outside the agent's reach, per-tool approval gates, model providers switchable without a redeploy, and a control plane for multi-tenant governance being built on the same architecture.

In Development — Present
06 Enterprise AI Platforms

PeopleForce MCP Server

An MCP server giving AI agents governed access to PeopleForce HR data: every tool carries a sensitivity tier (broad organizational data vs. recruiting and compensation), one codebase serves two deployments with different exposure profiles, and four independent enforcement layers fail closed. In early development, designed security-first before the first tool shipped.

In Development — Present
02 / 2 PROJECTS

Applied AI Products

AI products with real users, measured quality, and business outcomes.

03 / 10 PROJECTS

Open Source & Experiments

Public code and experiments: platform engineering depth anyone can read.

01 Open Source & Experiments

MCP Context Server

A high-performance Model Context Protocol server that gives LLM agents a durable, searchable memory layer beyond the context window. Thread-scoped storage over SQLite or PostgreSQL, with semantic, full-text, and hybrid search, cross-encoder reranking, pluggable embedding and summary providers, metadata filtering, and batch operations — all behind a standard MCP interface.

Active — Present
02 Open Source & Experiments

Claude Code Toolbox

A cross-platform installer and declarative environment-configuration framework for Claude Code. One YAML file captures all agents, MCP servers, slash commands, hooks, skills, rules, system prompts, and settings; one command installs everything on Windows, macOS, or Linux — portable, reproducible, centrally updatable.

Active — Present
03 Open Source & Experiments

AEGIS

An opinionated, hardened Claude Code environment that extends the harness rather than replacing it: curated skills, always-on rules, quality-enforcing hooks, MCP servers, durable cross-session memory, and an installed quality toolchain, with deliberate model, effort, and permission defaults. It rides on Claude Code's native dynamic-workflow orchestration and supplies the reliability and standards layer around it. Parts of it are extracted, generalized, and published in the Claude Code Artifacts library.

Active — Present
04 Open Source & Experiments

Claude Code Artifacts

A public, MIT-licensed library of reusable Claude Code artifacts — environment configurations, skills, rules, hooks, agents, and slash commands. Each artifact is a self-contained, environment-agnostic building block addressable by its raw URL: install one by referencing it from a single environment YAML that Claude Code Toolbox sets up for you, or copy it straight into your ~/.claude/ directory.

Active — Present
05 Open Source & Experiments

Agent Memory & Context Evaluation Benchmark

An open benchmark, in design, that measures whether structured persistent context improves multi-session agent task completion over no memory and file-based handoff — and at what cost in retrieval errors, latency, and tokens. Versioned dataset, held-out test set, repeated runs with confidence intervals, ablations, error taxonomy, published raw results.

Concept
06 Open Source & Experiments

CXR Draft Auditor

A research and educational quality-assurance tool that gives a chest X-ray draft impression a transparent second read. Two tiny 4B models — a fine-tuned MedGemma that grounds the image into labeled findings with bounding boxes and an NVIDIA Nemotron-3 Nano 4B that parses the draft into the same labels — feed a deterministic, model-free comparator that flags MISSING, UNSUPPORTED, and URGENT discrepancies, with every flag traced to a specific finding, a specific phrase, and a box on the image. Not a medical device and not a diagnostic tool.

Not Active
07 Open Source & Experiments

st-copy — Streamlit Copy-to-Clipboard Button Component

A tiny, theme-aware Streamlit component that adds a one-click copy-to-clipboard button to any Streamlit app. Distributed as the st-copy PyPI package; used in AI chat UIs and prompt-output surfaces where a clean, accessible copy button was missing from the Streamlit toolbox.

Active — Present
08 Open Source & Experiments

Hugo Artifacts

A public, MIT-licensed monorepo of reusable Hugo modules — style-agnostic content shortcodes alongside site-scale modules for the SEO head, client-side search, responsive images, an accessible carousel, social sharing, build-time Open Graph cards, the redirect map and URL manifest, the machine-readable layer an AI agent reads, and the Progressive Web App shell — each independently versioned with its own go.mod. Every artifact ships semantic BEM markup, data attributes, and zero CSS, so it drops into any number of unrelated sites and is styled entirely by the consumer.

Active — Present
09 Open Source & Experiments

Machine Learning Prep and Train Toolkit

The ML Prep and Train Toolkit is an integrated suite of Python scripts that automates the data analysis, preprocessing, and training phases for binary classification models — hyperparameter search and calibration per algorithm, a stacking ensemble over the base classifiers, an evaluation table that reports overfitting probability alongside the usual metrics, and a feature-importance or LIME explanation for every model. It simplifies and accelerates the machine learning pipeline for data scientists and ML engineers.

Not Active
10 Open Source & Experiments

Terraform Cloudflare Zone Module

This project encapsulates a Terraform module facilitating the efficient creation and management of Cloudflare zone resources, reducing the complexity of Cloudflare configurations. It is designed for DevOps engineers looking to streamline their workflow when working with Cloudflare within their infrastructure as code practices.

Not Active
04 / 5 PROJECTS

Sites & Digital Products

Complete digital products and the sites behind them, designed, built, and operated end to end.

01 Sites & Digital Products

alexfeel.info

This site, as a project: a custom editorial design system on Hugo and Tailwind v4, a full machine-readable layer for AI agents (Markdown twins, llms.txt, a one-fetch profile document), build-time Open Graph card composition, an enforced single-source identity record, a bank of validators gating every deploy — and a 404 page that runs a real o200k tokenizer in the browser.

Active — Present
02 Sites & Digital Products

Motion Bus Card Balance and Status Checks Telegram Bot

A Telegram bot designed to provide real-time balance and status checks for the Motion Bus Card, used for public transportation in Cyprus.

Active — Present
03 Sites & Digital Products

Motion Bus Card Bot Site

The companion website of the Motion Bus Card Checks Bot: a one-page landing in a custom Mediterranean civic design system that explains the free service and routes people to Telegram, plus a machine-readable layer that tells AI assistants exactly what the service can and cannot do — so an agent hands the user the right link instead of inventing an API.

Active — Present
04 Sites & Digital Products

Agile Uni

A bilingual (English/Russian) education platform for Agile methodologies, built and run solo: full two-language content parity, a free AI assistant as the flagship tool, paid courses with real commerce plumbing (marketplace feed syndication, Course structured data), and a portable Hugo build that deploys unchanged to four hosting platforms.

Active — Present
05 Sites & Digital Products

Single Platform for Retailers and Buyers

A 2023 concept for an AI-powered marketplace that put retailers and shoppers on one platform: fit matching from a body scan, a natural-language shopping assistant, and a virtual try-on, aimed squarely at the return rates that make fashion e-commerce expensive. Worked out end to end — product, economics, market entry — and pitched.

Concept