Observability / Tracing
Tracing-oriented tooling for LLM workflows to make model calls, execution paths, and system behavior easier to inspect during development and evaluation runs.
Tracing Observability LLM Systems
Work
selected systems + experiments
Applied AI Systems
RAG-focused systems work with an ops/debugging mindset: making retrieval and generation workflows easier to inspect, iterate, and operate beyond one-off prompt demos.
RAG Ops Backend
Evaluation / Experimentation
Experimental sandbox for LLM evaluation workflows, with a focus on testing prompts, model behavior, and verifier-style checks in a setup that is easier to compare and iterate on than ad hoc notebooks.
LLM Eval Testing Experimentation
GitHub Preview
No README media. Open repo for code, notes, and results.
Research / Modeling
Two-stage transformer work for stochastic cellular automata prediction with entropy-guided patching, focused on emergent dynamics and where token budget should be spent when local uncertainty increases.
Transformers CA Modeling Entropy Patching