SEBASTIAN SCHKUDLARA
PostgreSQL, Redis, TypeScript
Node.js, React
Multi-Agent Orchestration
LangChain, PydanticAI
Tool Calling & Routing, MCP
Prompt Engineering
Vector DBs (pgvector)
Event-Driven (Kafka)
K8s / AWS / GCP
Databases: MongoDB, Neo4j, MySQL, SQLite,
Memcached, Elasticsearch.
DevOps & Scripting: Bash/Shell, Ansible,
Terraform, GitHub Actions, Prometheus, Grafana.
AI Frameworks: LangChain, LangGraph, PydanticAI, Hugging Face, OpenAI API, Anthropic API, Gemini API.
Architecture: REST, GraphQL, gRPC, JSON-RPC, OAuth
2.0, WebSocket.
Blockchain (secondary): Polygon, Ethereum, Solidity,
Hyperledger, Hedera, Chainlink, Stellar.
- Polyglot: DE (Native), PL (Native), EN (Fluent), ES (Fluent).
- Hyper-Productivity: Rapid prototyping, TDD, & deployment.
- Team Buff: Mentorship, Team Building, & Technical Leadership.
Professional Summary
Senior Backend Architect turned Agentic AI Infrastructure specialist — 19+ years building distributed systems, now designing the tools and protocols that make autonomous AI agents work in production: routing proxies, P2P agent networks, A2A communication layers, and RAG pipelines engineered for scale.
Key Achievements: SwitchAI proxy reducing AI costs by 30% across multi-provider LLM routing · HiveMindAI cutting agent deployment from weeks to days · AI-driven logistics automation reducing manual processing by ~70% at ChainGO.
Stack: Kubernetes, AWS/GCP, FastAPI, LangChain, PydanticAI, RAG workflows, Python-based AI frameworks, and Ruby on Rails API development.
Open Source Projects
Open-source, decentralized protocol replacing traditional CVs with AI-agent-readable "Skill Cards." Candidates publish structured JSON/YAML/Markdown profiles to their own Git repos, making career data globally discoverable by any AI agent. Features a three-zone privacy model and anti-spam micropayments.
Unified AI gateway that bridges Gemini, Claude, Ollama, and CLI tools into a single OpenAI-compatible server. Features intelligent routing, load balancing, and autonomous self-healing with Superbrain intelligence.
Command-line interface for managing AI agents with conversational Cortex interface. Natural language processing, multi-turn conversations, and intelligent command execution.
P2P networking plugin for agent-to-agent communication. Features zero-trust identity with Ed25519, multi-transport support (NATS/libp2p), Circuit Relay v2 for NAT traversal, and comprehensive health monitoring.
Production-grade Ruby SDK for Google's Agent2Agent (A2A) Protocol. Enables seamless agent-to-agent communication via JSON-RPC 2.0, gRPC, and HTTP+JSON transports with full OAuth 2.0 support.
Creator | Engineered a high-fidelity HTML-to-Markdown conversion pipeline designed specifically for AI agents. Features deep site crawling, batch processing, and advanced WAF/Cloudflare bypass using stealth Puppeteer rotation. Includes a REST API for programmatic agent discovery and automatic "AI Skill Bundle" generation.
Enterprise-grade authentication clients for Traylinx Sentinel API with full A2A Protocol support. Available for Python and Node.js with OAuth 2.0, JWT, and secure token management.
GitHub Organizations:
@traylinx |
@rschumann
Additional projects include homebrew-traylinx, scoop-traylinx, a2a-ruby-samples, and google-drive-intelligent-mcp.
Core Competencies
- LLM Integration: OpenAI, Anthropic, Google, Mistral, LLaMA, DeepSeek, etc.
- RAG Pipelines: High-accuracy retrieval-augmented generation.
- AI Orchestration: PydanticAI, LangChain, custom multi-agent systems.
- Vector DBs: Pinecone, ChromaDB, pgvector.
- Cloud: Kubernetes, Docker, Terraform, AWS, GCP, Azure.
- Streaming: Kafka, RabbitMQ, Sidekiq, Redis Pub/Sub.
- CI/CD: GitHub Actions, Prometheus, Grafana.
- APIs: FastAPI (Python), Ruby on Rails, GraphQL, gRPC.
- Real-time: Redis, streaming analytics.
- Database: PostgreSQL, MongoDB, Neo4j - optimized for AI workloads.
Professional Experience
SwitchAI (Multi-Model AI Proxy): Architected and engineered an AI request-routing proxy, dynamically selecting models based on cost/latency. Unified integrations across OpenAI, Anthropic, Google, etc.
Scoutica Protocol: Designed an open-source protocol replacing CVs with AI-readable Skill Cards. Engineered the multi-agent recruitment platform underneath using event-driven AWS SNS/SQS.
HiveMindAI: Built a marketplace for autonomous AI agents. Developed Traylinx CLI for seamless agent lifecycle management.
2md (AI-Ready Markdown Engine): Engineered an advanced data ingestion engine designed specifically for AI agents and RAG pipelines. Converts any URL or document into structured, AI-ready Markdown. | 2md.traylinx.com
Advanced OAuth 2.0 Auth Service: Designed high-security microservice with specialized A2A flows. | Clients: Python, Node.js
A2A Ruby GEM: Developed production-grade SDK for Google's A2A protocol (gRPC, OAuth 2.0). | View Repo
Architected advanced RAG systems for industrial R&D. Led model fine-tuning initiatives for maritime and defense sectors.
Developed RAG-based AI solutions and optimized LLM performance for enterprise data infrastructure. Built NLP pipelines for summarization and sentiment analysis.
Pioneered AI-driven automation in logistics using microservices. Developed eBL Platform digitizing Bill of Lading via Polygon blockchain NFTs.
Grew team from 1 to 15 engineers. Established K8s infrastructure and built real-time event-driven analytics backend for insurance risk.
Developed large-scale environmental data APIs and unified diverse datasets via microservices for global research visibility.
Featured Projects Detail
Revolutionized the Bill of Lading process using Polygon blockchain to create NFT-backed eBLs, ensuring immutable ownership and traceability.
Blockchain-based system for forest product traceability, ensuring end-to-end transparency and compliance via blockchain certifications.
Developed backend APIs for a global environmental initiative (White House OSTP & WRI). Created scalable query services used by researchers worldwide.
Education & Expertise
TU Berlin
LLM fine-tuning, microservices best practices, and AI architectures.