https://caio.theodoro.dev/https://caio.theodoro.dev/about/https://caio.theodoro.dev/blog/https://caio.theodoro.dev/blog/building-benchmarks-like-arc-measuring-whether-it-works/https://caio.theodoro.dev/blog/calibration/https://caio.theodoro.dev/blog/economics-of-model-serving/https://caio.theodoro.dev/blog/evaluation-gates-reliability-moat/https://caio.theodoro.dev/blog/fine-tuning-when-it-makes-sense/https://caio.theodoro.dev/blog/llms-and-time-series-forecasting/https://caio.theodoro.dev/blog/mirofish-demand-forecasting/https://caio.theodoro.dev/blog/on-generative-ui/https://caio.theodoro.dev/blog/on-synthetic-data/https://caio.theodoro.dev/blog/ornith-curriculum-audit-distribution-wins/https://caio.theodoro.dev/blog/rag-retrieval-one-year-in/https://caio.theodoro.dev/blog/rag-vs-fine-tune-build-decision/https://caio.theodoro.dev/blog/reconforge-1-7b-beats-deepseek-on-the-money-metric/https://caio.theodoro.dev/blog/simulation-not-prediction/https://caio.theodoro.dev/blog/structured-outputs-llm-pipelines/https://caio.theodoro.dev/blog/suture-8b-three-gates-that-lied/https://caio.theodoro.dev/blog/thinking-with-visual-primitives/https://caio.theodoro.dev/blog/what-rlhf-actually-optimizes/https://caio.theodoro.dev/blog/what-superforecasters-got-right/https://caio.theodoro.dev/blog/what-the-graph-actually-adds/https://caio.theodoro.dev/blog/when-ai-agents-stop-taking-turns/https://caio.theodoro.dev/blog/why-lossbench-significant/https://caio.theodoro.dev/contact/https://caio.theodoro.dev/journey/https://caio.theodoro.dev/privacy/https://caio.theodoro.dev/projects/https://caio.theodoro.dev/projects/all/1/https://caio.theodoro.dev/projects/all/2/https://caio.theodoro.dev/projects/certifications/1/https://caio.theodoro.dev/projects/certifications/2/https://caio.theodoro.dev/projects/experience/1/https://caio.theodoro.dev/projects/experience/2/