I'm an independent researcher and engineer transitioning into full-time work in AI safety and alignment. I bring an interdisciplinary background in computational biology, computer science, philosophy, and systems engineering—with professional experience across data engineering, software development, and site reliability at scale.
🔭 Currently working on
- Prototype alignment environments for testing deceptive or misgeneralizing agents
- GaiaLogos: a research testbed for ecologically-bounded agent development
- Compression-based evaluation frameworks inspired by historiographical modeling
🌱 Currently learning
- Red/blue team evaluation techniques (
Inspect, alignment-faking games) - Recursive, epistemically-updating reward shaping
- Multi-objective reinforcement learning and scalable oversight design
👯 Looking to collaborate on
- Interpretability tools, deception detection, or oversight systems
- Evaluation frameworks that combine philosophical grounding and engineering fluency
- Alignment projects that treat AI as embedded in an ecological and civilizational context
🤔 Open to help with
- Understanding large-scale interpretability workflows (SAEs, tracing tools)
- Publishing or productionizing early-stage alignment tools
- Connecting with mentors or institutions working on practical safety architectures
💬 Ask me about
- Deceptive alignment, narrative compression, AI persuasion risks
- How ecology, ethics, and epistemology can be computationally encoded
- Philosophical perspectives on reward shaping and value alignment
📫 Reach me at
Twitter • LinkedIn • Email • Substack
😄 Pronouns: he/they
⚡ Fun fact: I'm working on reading through the top-100 books of all time. I'm currently on 48/100.