AI Agents & Autonomous Systems · Rust & Optimization · Consciousness Research · Ψ / Psi Research · Questioning Assumptions
Building at the intersection of intelligence, optimization, consciousness, and the occasional idea that probably should have remained a thought experiment.
I'm Aron Bijl, an AI builder, researcher, author, and experimental psychologist from the Netherlands.
I hold an MSc in Psychology from the University of Amsterdam, with a research background spanning experimental psychology, cognition, consciousness, and parapsychology.
My work tends to begin where disciplinary borders become inconvenient.
I build AI agents and infrastructure, develop unconventional optimization algorithms, investigate the architecture of intelligence and consciousness, and occasionally disappear down sufficiently deep theoretical rabbit holes that they turn into software, papers, books, or entirely new research programs.
My guiding principle is fairly simple:
Follow the evidence — especially when it goes somewhere inconvenient.
An experimental optimization engine exploring a geometric/topological approach to difficult search spaces.
Originally developed against combinatorial optimization problems, Pairwolf has since been pushed into increasingly different domains — from QAPlib and MaxSAT to EvoxBench and PGLibUC. A predator indeed.
The interesting question isn't merely whether an optimizer can find a solution.
It's whether there are deeper structures in problem spaces that conventional search methods are leaving unused.
Context engineering for increasingly long-lived AI agents.
Token Terminator explores aggressive context reduction without casually throwing away the information an agent may need later.
Current work includes:
- Semantic context selection
- Exact-recovery mechanisms
- Agent-specific context engines
- Token and inference-cost reduction
- JEV-based probabilistic decision layers
- Omission-risk estimation
- Persistent agent memory and context architecture
Because apparently giving an AI a gigantic context window only encourages it to fill the gigantic context window.
A voice-first point-of-sale system.
Speak the order. Done.
An experiment in replacing layers of touchscreen interaction with natural-language interfaces that actually save time in real operational environments.
I build and experiment with autonomous and semi-autonomous AI systems: tool-using agents, orchestrators, memory systems, context engines, verification layers, and multi-agent architectures.
I'm particularly interested in the point where an LLM stops being merely a chatbot and becomes a component inside a persistent cognitive system.
That raises engineering questions.
It also raises considerably stranger ones.
My academic background is in experimental psychology, particularly research near the boundaries of cognition and consciousness.
Areas that continue to interest me include:
Consciousness · Psi Research · Retrocausality · Cognition · Human–AI Interaction · Emergent Intelligence · Complex Systems
I've co-authored peer-reviewed research on retroactive effects and methodological questions in parapsychological research.
I remain stubbornly fond of hypotheses that can actually be tested.
One question increasingly connects almost everything I work on:
Not just how to make models score higher on benchmarks, but what computation, representation, memory, prediction, agency, and consciousness might have in common.
AI makes this question unusually difficult to avoid.
If increasingly sophisticated artificial systems begin exhibiting properties we once considered diagnostic of cognition, then we have two options:
- Keep moving the definition.
- Reconsider the definition.
I find option two more scientifically interesting.
My broader theoretical work also explores connections between information, geometry, optimization, physics, cognition, and complex systems.
Some of those ideas may be wrong.
That's what experiments are for.
I'm developing the Unified Matrix Model, an attempt to explore whether apparently separate structures across physics, information, geometry, cognition, and optimization might emerge from a deeper common framework.
It's deliberately ambitious.
I consider that a feature.
The project combines mathematical exploration, computational experimentation, pattern analysis, and a willingness to follow structural similarities before deciding whether they are profound or merely coincidental.
The objective isn't to defend a theory.
It's to build one that can eventually survive attempts to kill it.
Under the name Frater C.D.M., I wrote The Code of Reality, an exploration of the seven Hermetic principles through the lenses of psychology, systems thinking, consciousness, and practical experimentation.
My interests have never respected the supposed wall between the empirical and the esoteric.
I don't think either side benefits from refusing to look over it.
| Area | Things I work on |
|---|---|
| 🤖 AI / Agents | LLMs, agent orchestration, context engineering, RAG, autonomous workflows |
| 🧠 Cognitive Systems | Memory, verification, probabilistic control, context architecture |
| ⚙️ Optimization | Rust, combinatorial search, multi-objective optimization, unconventional search methods |
| 🔬 Research | Python, statistics, experimental design, data analysis |
| 👁️ Consciousness | Cognition, psi, anomalous phenomena, artificial consciousness |
| 🧬 Theory | Complex systems, information, geometry, physics, Unified Matrix Model |
| 💸 Current obsession | Making expensive intelligent systems considerably less wasteful |
- 📍 Netherlands
- 🧠 Experimental psychology
- ⚙️ AI systems & optimization
- 🔮 Consciousness and psi research
- 🐺 Pairwolf
- 🧬 Unified Matrix Model
- 🤖 Persistent AI agents
- ✍️ Writing when I should probably be sleeping
Want to talk AI, optimization, consciousness research, collaboration, consulting, research, or an interesting problem that doesn't fit neatly into a category?
I'm generally much easier to reach digitally than by phone. 😄
Particularly welcome:
- Research collaborations
- Interesting optimization problems
- AI and agent engineering
- Commercial applications of Pairwolf
- Context-engineering and inference-efficiency work
- Consciousness and psi research
- Serious interdisciplinary work
- Conversations that begin with "This may sound strange, but..."
If you're contacting me about a project, tell me what you're building, what the problem is, and why you think I might be the right person for it.
🌍 Website: aronbijl.com
💻 GitHub: @AronAxe
📧 Email: aron@bijl.eu
📖 The Code of Reality: thecodeofreality.com
If you came here looking for a normal collection of software projects, you may encounter:
AI agents → optimization → context engineering → consciousness → psi → geometry → physics → Hermeticism → "Wait, how did we get here?"
That's approximately how it happens to me too.
Everything since then has gotten slightly out of hand. ⚡