11 projects
nitelos
Verify the outcome, not the attempt: evidence-based success verification for AI agents.
locale-invoice-check
Matched-pair EN/DE/ES invoice extraction regression tests with deterministic CI gates.
vision-input-check
Regression tests for images sent to vision LLM APIs: pixel-identical and lossy delivery variants, variance-aware scoring, CI gates.
voice-evals
Evaluation harness for voice agents: replay scoring plus a scripted live probe with WER, latency budgets, barge-in behavior and task outcomes.
clef-router
OpenAI-compatible proxy that routes prompts between cheap and frontier LLMs using Cloudflare's Clef decision model
clef-compactor
Query-aware RAG context compaction using Cloudflare's Clef. Keep the evidence, cut the noise.
clef-evals
Calibration-first evaluation toolkit for Cloudflare's Clef decision models. Judge cheap, audit confidence.
laya-evals
LLM-as-judge and calibration auditing for System 1 decision models: rubric scoring in CI plus reliability diagrams, ECE and coverage/accuracy thresholds by option count
laya-phishield
Explainable phishing email detection: atomic semantic signals from a local System 1 decision model plus deterministic header/URL checks, combined by logistic regression
laya-compactor
Context compaction for RAG and agents: relevance-score every chunk with a local System 1 decision model and cut tokens before they reach the LLM
laya-router
OpenAI-compatible proxy that routes prompts to cheap or frontier LLMs using a local System 1 decision model (laya)