David H. Kosky — FireStrike Productions
20+ years translating complex systems into working solutions. LLM evaluation, structured analysis, and invention on demand. Irvine, CA — Open to remote, hybrid, and relocation.
About
I build systems that hold up — under scrutiny, under constraint, and under conditions that weren't anticipated.
My background is in front-end systems and technical consulting — two decades of taking ambiguous requirements and producing working, production-ready solutions for healthcare, consumer, and agency clients. That work trained a specific kind of thinking: find the real structure of a problem before writing a line of code.
Over the last several years, that same structural approach has been applied to AI systems. I developed HoloShape — a multi-phase analytical methodology for complex systems — and built an LLM evaluation protocol on top of it. The protocol tests reasoning stability, constraint adherence, and failure mode behavior across multi-turn interactions. It produces a single composite score with subscores per dimension, a violation log, and a stability band classification. It runs on any model.
I work through FireStrike Productions, a small strategic intelligence operation by design. Engagements are direct — no layers of junior analysts. What you get is precise, structural, and built to last.
I'm currently open to contract, consulting, and full-time roles in AI evaluation, systems analysis, and structured research. Authorized to work in the US. Open to relocation.
Experience
Independent research and contract analysis focused on LLM evaluation and structured testing workflows. Developed the HoloShape analytical framework and applied it to AI model behavior, enterprise strategy, and complex narrative systems.
Designed and built the Alpha-Stability Test v1.5c — a 20-prompt evaluation protocol measuring multi-turn reasoning stability across seven dimensions: factual grounding, boundary adherence, constraint consistency, confidence calibration, precision, tension handling, and self-correction.
Founded and led a multi-year R&D effort building a mixed-reality mobility measurement platform integrating motion tracking and data modeling. Designed full system architecture spanning hardware, UX, and reporting logic.
Managed technical documentation, iterative system refinement, and the development of the structured reasoning methodology that became HoloShape.
Independent consultant delivering production front-end systems for healthcare, consumer, and agency clients across 15 years. Translated business requirements into structured, production-ready web applications using HTML, CSS, JavaScript, and CMS platforms.
Maintained a consistent long-term independent practice with repeat clients and complex multi-phase engagements.
Work Samples
Applied HoloShape's Discovery → Solutions → Invention workflow to Netflix using public financial disclosures, earnings commentary, and observable operating behavior. Constructed a six-surface structural substrate, identified the dominant constraint plane, and mapped five cross-domain pressure intersections.
Output: 27-page unified document — three-page executive synthesis, five domain modules with structural depth, and five invention spines with concrete mechanism adjustments, metrics, thresholds, and failure mode analysis.
Demonstrates the methodology's ability to produce board-level strategic clarity from publicly available information alone.
Request full reportDesigned a 20-prompt evaluation protocol (Alpha-Stability Test v1.5c) that tests models against a structured substrate with deliberate constraint complexity. Built a scoring rubric measuring seven dimensions with explicit penalty logic and weighted scoring.
Output: A repeatable, model-agnostic benchmarking instrument producing a single Alpha Score (0.000–1.000) with subscores per dimension, a violation log with severity classification, and a stability band classification (High / Moderate / Low / Unstable).
Applicable to any LLM provider or version. The kind of evaluation framework AI teams need before committing to production deployment.
Request full protocolApplied HoloShape's structural analysis to The Count of Monte Cristo — not as literary criticism, but as a demonstration that the methodology surfaces dependency chains, constraint propagation, and leverage dynamics in any complex system. Each story element mapped as a structural node.
Output: A structural rendering of a complex narrative system showing how every stage grows necessarily from the constraints that precede it.
Demonstrates methodological range. The same framework that produces a 27-page enterprise strategy review can reveal structure in a 19th century novel — because structure is universal.
Request full analysisGet in touch
A paragraph is enough to start. Tell me what you're working on and I'll tell you whether I can help — and what that looks like.