csauer.devFor Anthony Traniello / MD IT, CISO
01 / A Proposal

Chris Sauer

IT Project Manager / AI Software Developer

Testifying expert. Builds the AI tooling. Defensible against cross-examination.

Chris Sauer
02 / The Ask

Senior AI engineering with deep domain expertise built in, without an outside hire.

Half my time stays billable in TCA. The other half is corporate. AI tooling for our billable work, built by someone who'll defend it under cross-examination.

03 / Responsibilities
Execution
End-to-end: scope, build, documentation, rollout, support.
User Advocacy
User advocate and product engineer. Builds tools for our experts who use them — with deep domain knowledge of what's actually useful for our billable matters. High agency, defined taste.
AI Development
Production pipelines with built-in trust verification — every output traceable to source, aligned with what testimony requires.
HW / SW Lifecycle
Inventory, replacement planning, entitlement, vendors.
Communication
Between users, leadership, and IT teams.
04 / Position on AI in Expert Work
We use AI to help us do the work. We don't use AI to do the analysis. The opinions are ours.
05 / Why Chris Sauer

The role pairs production AI engineering with forensic domain expertise and the credibility to defend the outputs in deposition.

Twenty-two years in the construction industry, eight at SOCOTEC. Mechanical engineering, Georgia Tech. Expert testimony in AAA arbitration and mediation: Marin Health Center ($535M), the Pecém ($4.5B) and ThyssenKrupp CSA (€6B) steel mills, Aleris Lewisport. DRB support on Vogtle Units 3 & 4 ($25B). Moderated the AI for Testifiers panel at the 2026 SOCOTEC Testifiers Summit.

22
Years in Forensics
8
Years at SOCOTEC
5
Production Tools Shipped
2
Office Buildouts
06 / Already Shipping

Projects delivered.

All detail shown — click any project to collapse.

Production methodology applied across multiple billable matters. Two examples below.

AI matches client lane-closure records to DOT base-map geometry. Date-ranged dashboard renders every result on the base map beside its source record.

Production computer vision pipeline. Staged AI analysis classifies workers as masked / unmasked / indeterminate, annotated against the original image.

Owned scope, directed delivery, authored the user guide and IT implementation checklist. Rolled out across Advisory East. Used daily, low IT ticket volume.

P6 export utility paired with Tableau workbooks for schedule analysis. Built AI-assisted with limited external developer help.

On time, on budget. Integrated conference-room systems that work — no IT tickets.

Built and ran the original framework.

07 / Working Relationships

Translation layer between corporate IT/AI capability and the Advisory and TCA practices. Partners with Anthony Traniello on engineering and AI strategy. Coordinates with practice leadership on matter-driven requirements.

08 / How the Work Gets Built

Agentic AI engineering.

Work decomposed into a Dolt-versioned issue ledger ("beads") and executed by autonomous AI engineering agents — each spawned into its own isolated git worktree, gated by an automated merge queue. The unit of work lives immutably in the SQL bead, so even if an agent dies mid-stream, the work continues on respawn.

Stack
Data
Postgres · Supabase · PostGIS
AI
OpenAI · Anthropic · AWS Textract
Workflow
n8n · Gastown
App
Next.js · TypeScript · Python
Deploy
Vercel · Railway · Hetzner
Versioning
Dolt · GitHub CI/CD