Everything here is free, with no email wall. Read the research portfolio, work through the diligence checklists online, print them, or take the PDF.
AI Research and Systems Portfolio
Cisco Caceres’s research and engineering contributions, project evidence, completed experiments and next studies. A dated portfolio for technical review.
Technical Due Diligence Checklist
110+ checks across 11 sections, from the five-minute smell test to licensing, AI-generated code and the final call. Run on real diligence engagements.
AI Code Provenance Audit
How much of this codebase was generated, by what, reviewed by whom, and can anyone fix it at 2am. The question most diligence checklists miss.
Founder's Pre-Diligence Self-Audit
What an investor will find in your codebase before they find it. The checks I run on a target, pointed at you, while there is still time to fix things.
The Rewrite Decision
Rewrite or refactor, and how to tell which one you're facing. A scored framework, plus the conditions under which rewrites actually succeed.
Cloud Cost Reality Check
What your infrastructure actually costs per user and per transaction, where the waste hides, and which savings are real rather than a spreadsheet exercise.
Verifier Information & Missing-Outcome Bounds
Elementary decision-theoretic ceilings and fixed-bank bounds for missing execution outcomes, with explicit limits for learned controllers and paid verification.
Machine Agency Infrastructure
An architecture draft for reliable machine agency, distinguishing implemented interfaces, proposed connections, evidence contracts and publication gates.
Verifier Reliability Under Budgeted Code Generation: Error Covariance and Information-Theoretic Stopping Bounds
Theoretical stopping bounds and empirical verification calibration under positive error covariance, contrasting naive Best-of-N against covariance-aware sequential allocation across multi-budget sweeps.
Tactical Radio Channel Simulator (TRCS): Characterizing Non-Speech Audio Hallucination in Speech Foundation Models
Controlled empirical evaluation of non-speech audio hallucination and narrowband telephone codec degradation across six open speech models and 30 channel conditions.
More from the Field Manual
The deeper writing lives on RoamingPigs, an independent technical publication I edit and publish. 170+ articles. Free. RSS and weekly Sunday digest.