Third City Analytics
Legible systems built to discern, execute, and learn.
Scroll to exploreCapabilities
01
We develop methods for layered, heterogeneous graph structures where the most analytically interesting entities resist clean classification — nodes with probabilistic membership across multiple communities simultaneously. Our work covers convergence frameworks for multi-layer graphs using graphon theory and stochastic block models, taxonomy and classification of ambiguous node behavior, and behavior-based traversal algorithms for efficiently navigating complex relational networks. Applications include financial network intelligence, supply chain risk, and entity resolution across disparate data systems.
02
We turn physical environments into structured, queryable datasets — condition assessments, compliance scoring, and capital prioritization across large property portfolios. Computer vision pipelines operating at scale, with traceable visual evidence behind every finding.
03
Orchestrated AI systems for analytically complex, multi-stage workflows. We design agent hierarchies that decompose hard problems, enforce output quality, and synthesize across parallel workstreams — producing conclusions, not just data.
04
Econometric and formal modeling where clean experimental designs are unavailable. Agent-based simulation, quasi-experimental methods, and formal system specification under incomplete information.
05
Analytical platforms for environments where the data picture is incomplete, rapidly evolving, and the cost of delay compounds. Our work in this area centers on catastrophe-affected environments — building systems that accelerate situational understanding, triage, and coordination when conventional infrastructure has degraded. The goal is to compress the gap between an event and a legible picture of it, so that recovery and resource allocation decisions can be made on evidence rather than assumption.
If you have a problem where the standard toolbox does not reach, we would like to hear about it.