Senior Technology Leader | 25 Years Experience | CTO / Fractional CTO / Senior C# Developer

Building scalable solutions and leading high-performance teams

I'm a senior technology leader with 25 years of development experience, 15+ years building scalable applications and 7+ years leading international development teams. I transform technology organizations from startup-phase to enterprise-grade while achieving market leadership positions.

My approach: Remove bottlenecks and dependencies for sustainable growth—both in code architecture and team dynamics.

• Senior C# Developer positions
• Technical Architecture consulting

Focus Areas: FinTech, RegTech, Enterprise SaaS, Start-ups needing technical foundation, AI/ML applications

Location: Amsterdam area (Weesp, Naarden, Almere, Bussum, Hilversum) or remote/hybrid

AI-powered energy management for ME/CFS patients and cancer survivors with mitochondrial dysfunction

(Personal Energy Monitor, Predictor & Analytics Logger)

Tech Stack: Rust, Deno, Burn (ML), DuckDB-WASM, LSTM/TFT models

What it does: Privacy-first health analytics that discovers personal crash patterns from wearable data. Predicts Post-Exertional Malaise (PEM) risk using HRV, heart rate, sleep, and activity metrics—helping users avoid energy crashes that can last days or weeks.

Who it helps:

• ME/CFS patients - Manage energy envelopes and prevent debilitating crashes
• Cancer survivors - Navigate recovery with mitochondrial damage from chemotherapy/radiation
• Long COVID patients - Understand post-viral energy patterns and pacing

Business Value: Personalized ML models identify individual warning signs and safe activity levels. All data processed locally in the browser for complete privacy—no health data leaves the user's device.

Self-improving multi-agent orchestration system with NER and knowledge graph capabilities

Tech Stack: Rust, libSQL/Turso, GLiNER, Tree-sitter, MCP Server

What it does: Enables AI agents to coordinate as a swarm using stigmergy (environment-based coordination), eliminating context bloat while enabling massive parallelization. Self-improving system that learns optimal execution strategies over time.

Key Crates:

• fab-swarm - Core CLI and MCP server with 40+ tools, self-healing swarm orchestration
• fab-brain - Personal knowledge graph with semantic search, "second brain" for AI-assisted knowledge curation
• fab-learn - Learning system that tracks outcomes, learns optimal tier routing over time
• fab-entity - High-performance graph-based NER with GLiNER (NAACL 2024 SOTA zero-shot entity extraction)
• fab-codebox - Tree-sitter AST cache with query REPL for code intelligence (40+ languages)
• fab-lint - Fast technical debt linter with jj integration

What Makes It Different:

• Nano-agents: Deterministic tasks execute in ~50-500μs without LLM calls (vs. seconds for traditional agents)
• Stigmergy over messaging: Agents coordinate through shared environment, not message passing (avoids N² message explosion)
• Auto-improving: fab-learn tracks which model tiers work best for each task type, continuously optimizing
• Self-healing: Automatic recovery from crashed agents with zero downtime
• Production-ready: Follows hexagonal architecture, comprehensive testing with cargo-llvm-cov (44% coverage)

Practices Followed:

• Hexagonal architecture with clean separation of concerns
• Dependency injection for testability
• Circuit breaker pattern for external API resilience
• Rate limit pooling for distributed systems
• Graceful shutdown with LIFO cleanup ordering

Performance analyzers for C# 11-14 and .NET 9-10

Tech Stack: C#, Roslyn Analyzers, .NET 9-10 RC1

What it does: 23 Roslyn analyzers providing 10-200x performance improvements for C# codebases

Business Value: Catches performance anti-patterns at compile-time before they reach production, eliminating costly refactoring cycles

Complete C# 14 implementation of Google's ReasoningBank with self-learning capabilities

Tech Stack: C# 14, .NET 10, TensorPrimitives, SIMD (AVX-512/ARM SVE), Microsoft Agent Framework

What it does: Production-ready AI memory engine built on Google's ReasoningBank architecture, fully implemented in C# 14 with .NET 10's native TensorPrimitives and SIMD vectorization for 30-250x performance improvements.

Self-Learning Capabilities:

• 9 Modern RL Algorithms: MCTS, DQN, Dueling DQN, Rainbow, A2C, PPO, SAC, Q-Learning, Multi-Agent systems
• Reflexion Memory: Self-critique and learning from failures
• Skill Library: Pattern consolidation with k-means clustering
• Causal Memory Graph: Pearl's do-calculus for understanding cause-and-effect
• Propensity Score Methods: IPW (Inverse Probability Weighting) for causal inference

Key Features:

• Hardware-accelerated vector operations with TensorPrimitives (AVX-512/ARM SVE)
• Zero-allocation hot paths with Span for maximum performance
• Microsoft Agent Framework integration for AI workflow orchestration
• 340+ tests with 90-95% coverage
• 20 NuGet packages ready for production
• Docker support with multi-stage builds

Performance:

• 30-250x faster than JavaScript implementation
• Batch operations: 50,000 vector inserts/sec, 100,000 deletes/sec
• 75% memory reduction compared to Node.js (200MB vs 800MB for 1M vectors)

Status: 100% complete, production-ready, fully documented with 25+ guides

Code graph analysis and visualization tool

Tech Stack: Rust, Graph Algorithms, Matryoshka Embeddings

What it does: Analyzes codebases as graphs to identify patterns, dependencies, and architectural structure. Multi-language support (Rust, Python, TypeScript, C#) with incremental parsing that only re-scans changed files.

Key Features:

• Graph-based code analysis - Transform codebases into queryable dependency graphs
• Multi-language parsing - Support for 4+ languages
• Incremental scanning - Only re-parse modified files for performance
• Claude Code integration - MCP server for AI-assisted code exploration

• Fractional CTO - Helped startups scale from MVP to enterprise-grade architecture, led international teams through critical transformation phases
• Technical Lead - Architected FinTech, RegTech, and SaaS solutions

• Email: [Available upon request]
• Location: Amsterdam area, Netherlands
• Open To: New opportunities, connections, and collaborations