Anthropic Fellows Application

A field-tested operator who learned AI systems by building them.

I am not presenting myself as a lifelong software engineer. I am a lifelong builder: construction founder, Navy veteran, superintendent, project manager, operational excellence leader, Lean coach, and now a self-taught AI systems builder using modern foundation models to turn ambiguous problems into working tools.

Honest Origin 30+ years building people, projects, processes, and reliability in the physical world.
Recent AI Growth Learned AI-assisted development rapidly through APIs, agent workflows, Python/JS scripting, and real users.
Research Motive Built public learning labs to understand model capability, safety behavior, and accountability.
WatchLocal.ai

WatchLocal Growth Loop

25 agents organized into 11 operating groups, driving 70 recurring actions across search, answer engines, reviews, listings, websites, content, competitors, and AI recommendations.

Growth Loop
Runs 24/7
Xavier
πŸ‘”
Scout
πŸ”
Forge
πŸ”§
Lens
πŸ“Š
Atlas
πŸ—ΊοΈ
Sentinel
πŸ›‘οΈ
Beacon
πŸ“‘
Pulse
πŸ’“
Bridge
πŸŒ‰
Clock
⏰
Amplify
πŸ“£
25 agents

Proof Projects

Three artifacts that show how I learn: build the system, publish the method, let real users inspect the result.

WatchLocal.ai LLC

Customer-facing company

Texas veteran-owned LLC providing AI-era visibility services for local small businesses across SEO, AEO, and GEO.

81 platforms 25 agents OAuth reports
What this proves Customer empathy, agentic workflow design, practical automation, and clear operating constraints. Open WatchLocal.ai

Training Run

Public AI benchmark learning lab

Built to understand the model ecosystem, compare hundreds of models, and make benchmark claims easier to inspect.

37 sources 10 pillars daily data methodology
What this proves Empirical curiosity, benchmark literacy, source discipline, and transparent scoring design. Open Training Run

The Safety Arena

Public AI safety learning lab

Safety-only model ranking concept using blind battles, curated prompts, human votes, and anti-gaming architecture.

blind battles safety prompts vote integrity audit logs
What this proves Alignment motivation, adversarial thinking, public accountability, and human-centered evaluation design. Open TS Arena

Evidence Map

Selected signals matched to Fellows criteria: implement quickly, communicate clearly, and work on empirical AI projects.

Built Working Systems

  • WatchLocal.ai autonomous local visibility workflow.
  • Training Run benchmark aggregation and public methodology.
  • Safety Arena blind battle / vote integrity concept.

Practical Technical Stack

  • Python and JavaScript scripting, HTML/CSS, APIs, JSON, Markdown.
  • Supabase, PostgreSQL, Edge Functions, RLS, JWT, OAuth.
  • Telegram and agent communication workflows.

Builder Timeline

The through-line is systems: physical construction, operational excellence, then AI-assisted tools.

1988-1994
United States Navy

Signalman / Customs Officer; high-stakes communication and Gulf War / Operation Desert Shield service.

1995-2005
Owner & Founder

Grew a residential and light commercial construction company with up to 14 team members.

2012-Present
Turner Construction

Project Manager, then Operational Excellence Manager: Daily Management Systems, A3, Lean, leadership coaching, and practical reliability.

2026
WatchLocal.ai, Training Run, Safety Arena

Self-directed AI systems learning turned into public proof: customer-facing automation, model benchmarking, and safety evaluation concepts.