l13v is a small AI research lab performing applied research across agent harnessing, ontology-based query systems, computer vision, and nonlinear signal processing — the convergence of a maths and physics foundation, complexity science (emergence, chaos, nonlinear dynamics), vision and control systems, and a decade of high-precision software engineering.
Independent evaluation of semantic query layers against raw-backend ground truth: every claim executed twice, four failure classes, an S1–S4 severity scale, and a landing pad for the correctness reports filed upstream.
→ detailsFrameworks and methodologies for human–agent collaboration in software engineering: multi-phase development workflows, engineering-standards enforcement via agents, and long-horizon memory and retrieval for agent sessions, with evaluation of agent reliability.
Domain-agnostic ontology inference from heterogeneous structured and unstructured data sources, natural-language to structured-query translation, and benchmark-driven evaluation of query-generation quality.
AI-driven sports-video understanding: player and ball tracking, event detection, and fan-experience enhancement.
Analysis of nonlinear and chaotic systems via higher-order statistics and nonlinear time-series methods. Originated in partial-discharge research in high-voltage insulation; the methods recur across signal-processing problems.
A family of frameworks for AI-augmented development: a multi-phase engineering methodology for human–agent collaboration, a mechanism for authoring and enforcing engineering standards through agents, and an advanced memory-retrieval layer for long-running agent sessions.
Tracking systems for automated sports-video analysis: player and ball tracking, event detection, and game understanding. Ongoing research into AI-driven sports-video understanding and fan-experience enhancement.
Full texts behind the DOIs on IEEE Xplore; reprints on request.