Model architecture
Sequence representation, contextual embeddings, training objectives, and scoring behavior.
Continue the conversation
See the implementation behind the public story.
The public site explains the problem, the behavioral-sequence thesis, and a set of synthetic demonstrations. A private screen-share goes deeper into the system design, implementation decisions, evaluation, analyst workflow, and the work still required to take Narrative beyond a proof of concept.
Walkthrough format
- Format
- Live screen-share and discussion
- Typical length
- 30–45 minutes
- Scope
- Tailored to the role, technical problem, or collaboration under discussion
- Data
- Public datasets and synthetic scenarios only—no client or employer information
Available in the walkthrough
Technical depth, with the tradeoffs left intact.
This is not a polished sales demo. The walkthrough includes the working system, design rationale, current limitations, and the decisions that would shape a production version.
Evaluation
Public-dataset experiments, calibration, failure analysis, and the limits of the current evidence.
Security data pipeline
Event contracts, OCSF normalization, tokenization, user context, and peer-group modeling.
Edge runtime
Quantization, inference constraints, state management, deployment choices, and tradeoffs.
Analyst experience
Alert triage, investigation context, user timelines, model review, and red-team analysis.
Production roadmap
Known integration gaps, security hardening, operational controls, and the path beyond the POC.
A useful conversation when
There is a real technical question to explore.
- Technical and architecture interviews
- Security or AI leadership discussions
- Potential collaborators and advisors
- Serious product or consulting conversations
Narrative Security
Bring the role, the problem, and the decision criteria.
I'll tailor the discussion to the aspects of the project that are most useful for evaluating fit, technical judgment, and implementation depth.