Building audit-grade AI agents and evaluation tooling in the open. Every number below regenerates from a committed script at github.com/ebt55. • diffing-agent-bench: sealed, preregistered benchmark for black-box model-diffing agents. Five LoRA finetunes of Qwen3.5-9B, answer key committed before any run, audited blind by a published recipe and four cheaper conditions. The agent asked the decisive database question in 0 of 13 attempts; a $0.15 prompt battery found the plant against $3.14 per agent run. MATS 12.0 work sample. • odd-number-forensics: forensic study of a published reward-hacking environment, ~5,300 audited samples. One model swings 0% to 87% gaming on single-line prompt edits. Falsified predictions published with the confirmed ones. • ProofPack: pre-approval review agent for Medicaid-audited disability programs. Every "Found" cites a SHA-256-hashed screenshot, quotes must match page text verbatim, and hashes are written by code the model never touches. $0.02–$0.19 per review, ~3 min vs 20–40 by hand, 87 offline tests. • IncidentGate: lab for policy gates, an action monitor and single-use human-approval tokens over an incident agent. 594 kill points, 434/434 recoveries, zero duplicate mutations. Closed at a baseline Sep 2026; on one scenario the full chain matched no safeguards at all, 12 side effects. • Digital Grimace Scale: preregistered Apart Research study. Primary test failed and is published as such; a second frozen plan found an effect (p = 0.005) that DPO training removed from the language while the behaviour stayed. • ExactDoc: PDF to editable DOCX, verified by render-back and word-position diffs. 16/16 corpus, 95.9% text kept live, 99 unit + 83 gate-mutation tests. • whose-voice: blind attribution of hidden principals in poisoned training data (Apart × Formation hackathon). 12–44% top-1 of 47 candidates, chance 2.1%, with a map of where the method breaks.

Worked as a Python backend engineer at a service company building MVPs, prototypes, data tools, and applied AI systems for startup clients across the US, Canada, Europe, and Australia. Core responsibilities: • Built FastAPI/Python backends for AI, data, and workflow automation products. • Designed backend architecture for 0→1 MVPs: API structure, data schemas, integrations, and deployment flows. • Integrated LLM APIs, vector databases, speech APIs, cloud services, and data pipelines. • Worked directly with ambiguous client requirements and translated them into functioning backend systems. Selected projects: • Speech assessment backend: processed short candidate videos using Whisper and Microsoft Pronunciation APIs; deployed on AWS Fargate. • Natural-language search: built an LLM-to-SQL backend for real-estate search using GPT-3.5, Cloud SQL, and GCP Cloud Run. • Document search/RAG: built ingestion and vector search flows using Qdrant over 1000+ documents. • Healthcare/device tools: built Python scripts and small GUI/debug tools for healthcare/device-adjacent projects. • IoT data handler: consumed MQTT streams and inserted live sensor data into MongoDB. • Report generation: built a Python API to generate complex PDF reports with nested tables and diagram images. • LLM infra prototype: fine-tuned Mistral 7B and served inference using vLLM/GKE as a prototype/learning project. Tech: Python, FastAPI, Docker, PostgreSQL, MongoDB, SQL, Qdrant, OpenAI APIs, Whisper, AWS Fargate, GCP Cloud Run, GitHub Actions, pytest.