Independent senior AI infrastructure and backend engineer | Germany / Spain / Poland / EU
I design and harden production AI systems: multi-agent orchestration, secure model gateways, RAG infrastructure, evaluation pipelines, and governance controls.
I bring 19+ years of backend and distributed-systems experience, with current work focused on making AI systems secure, reproducible, and operational.
Available for scoped B2B architecture, implementation, audits, and production hardening.
Creator and lead architect. A local-first operating layer that gives AI CLIs shared memory, search, sandboxed plugins, MCP tools, and cross-session automation across macOS, Linux, and Windows.
Relevant areas: agent infrastructure, permissions, persistent knowledge, plugin systems, Python, Rust, MCP, SQLite/FTS5.
Creator. A validator-in-the-loop workflow for multi-phase implementation, independent cross-model review, evidence gates, SARIF output, and recurring-finding memory.
Relevant areas: AI evaluation, reproducible execution, audit trails, multi-agent coordination, Python, CLI tooling.
Creator and lead architect. A Go-based OpenAI-compatible gateway that provides a common interface over multiple model providers and local AI backends.
Relevant areas: model routing, provider abstraction, API compatibility, gateway reliability, Go, observability.
Creator and lead architect. Public-beta tooling for private AI pods, WireGuard-isolated connectivity, a stable OpenAI-compatible gateway, shared files, and a local desktop environment.
Relevant areas: AI infrastructure, isolation, networking, gateways, Rust, Python, WireGuard, S3-compatible storage.
Creator. An open-source web-to-Markdown ingestion system for AI agents and RAG workflows, with JavaScript rendering, batch conversion, site crawling, sitemap discovery, and agent-ready output.
Relevant areas: content extraction, web crawling, RAG ingestion, Puppeteer, Node.js, Express, Redis, and queue-based processing. Live service.
Creator. An open protocol, schemas, and CLI for user-owned skill cards, deterministic opportunity matching, and auditable agent-mediated recruiting workflows.
Relevant areas: agent interoperability, JSON Schema, privacy boundaries, deterministic scoring, protocol design.
Creator and maintainer. A Ruby SDK for Google's Agent2Agent protocol with JSON-RPC, gRPC, and HTTP+JSON transports.
Relevant areas: agent-to-agent communication, SDK design, Ruby, transport abstraction, OAuth 2.0.
-
Traylinx — Private AI infrastructure and control surface for pods, gateways, shared folders, agents, API access, and account management.
-
Tytus OS — Public-beta local desktop and CLI for private AI pods, WireGuard connectivity, an OpenAI-compatible gateway, shared files, and agent applications.
-
2md — Converts web pages and sites into clean Markdown for AI agents and RAG workflows, including JavaScript-rendered pages, crawling, and sitemap discovery.
-
Makakoo — Open-source local-first operating layer that gives AI CLIs shared memory, search, tools, plugins, and cross-session context.
-
Traylinx Agentic Search — OpenAI-compatible multi-source web search and URL extraction with reranking, scraping, and cited summaries.
- Secure AI infrastructure and OpenAI-compatible gateways
- Multi-agent orchestration, evaluation, and governance
- RAG pipelines, model artifacts, and reproducible experiments
- Distributed systems, APIs, packaging, CI/CD, and production hardening
- Python, Rust, Go, Ruby, TypeScript, PostgreSQL, Redis, Docker, Kubernetes, AWS, and GCP
- Independent B2B engagements across Germany, Spain, Poland, the wider EU, and remote international teams
- Architecture, implementation, technical audits, and production hardening
- Written-first communication and evidence-based delivery
- German and Polish native; English and Spanish fluent
- Email: sebastian.schkudlara@gmail.com
- LinkedIn: linkedin.com/in/schkudlara
- Portfolio: jevvellabs.com/rschumann