Public claim
The portfolio shows aggregate investment scale and the Pittsburgh slice while keeping source-restricted firm-level tables private.
Pablo Zavala · AI Safety Evaluation · Research Engineering
AI Investment Mapping, Block Center, CMU
A Block Center project mapping more than ten billion dollars in public and private AI investment across eleven metropolitan economies for regional AI-readiness research. The Pittsburgh slice covers 6.3 billion dollars across 133 firms.
$10B+ mapped across 11 metros; Pittsburgh: $6.3B across 133 firms
Company-level records and maps stay private; the public page shows aggregates and methods evidence.
Role: Research and data pipeline contributor at CMU's Block Center.
How to Inspect This Work
Pipeline evidence
The public RAG evaluation shows retrieval-evaluation discipline; it does not verify the restricted investment dataset.
Boundary
A reader should treat this as a public summary of restricted research materials, not as a release of the underlying dataset.
Case Study
Problem
Regional AI policy often talks about national totals, while investment actually lands through specific metros, firms, and institutions.
Setup
The project builds a structured map of public and private AI investment across eleven metropolitan economies, including a deeper Pittsburgh slice.
Method
I contributed to data engineering and RAG-supported evidence review, with company and regional records kept in source-restricted materials.
Result
The research maps more than ten billion dollars in AI investment, including 6.3 billion dollars across 133 Pittsburgh firms.
Limitation
Company-level tables and maps stay private because the source data is restricted. The public page uses aggregate evidence only.
Evidence
The public RAG evaluation harness documents retrieval-evaluation discipline; it does not verify the restricted investment totals.
Key Outcomes
- More than ten billion dollars in AI investment mapped across eleven metropolitan areas
- Pittsburgh mapped at 6.3 billion dollars across 133 firms
- Public RAG harness demonstrates retrieval-evaluation discipline; it does not verify restricted investment totals
Methods
- Retrieval-augmented generation
- LLM pipelines
- Data engineering