Three focused areas: LLM application security, mobile device forensics, and cryptocurrency investigation. Each handled with experienced, hands-on expertise and clear accountability from first scope to signed report. Project-based engagements with written deliverables.
Think like an attacker. Write like an expert witness. Ship like an engineer.
Forensic and adversarial work for the people who need it most, from a practice that can build the tool and break the system.
Secure Data Consortium is an independent cybersecurity practice based in Jacksonville, Florida. The practice combines over a decade building production systems across financial services, government, and security tooling with active adversarial research and digital forensics.
Most LLM-adjacent consulting comes from one of two directions: security generalists who don't ship code, or developers who don't think about security architecture. This practice is both: builder-architect expertise paired with CISSP-level security depth.
Every engagement is project-shaped, handled by experienced, hands-on expertise from start to finish, and delivered as a written report.
More About the Practice →Deliberately narrow. Each area rewards experienced, hands-on expertise that has actually built and broken the systems in question.
Custom LLM application builds, on-premises model deployment, RAG pipelines for regulated corpora, and adversarial testing, including EU AI Act adversarial-testing evidence and indirect prompt injection assessment.
Learn more →Consent-based Android and iOS forensic examination for family-law attorneys, domestic-violence organizations, and civil matters. Stalkerware detection. Deposition-grade written reports.
Learn more →Blockchain forensic tracing for scam victims, civil litigation, and attorneys pursuing recovery. Documented methodology, honest scope. Former Chainalysis engineer; FLETC co-instructor.
Learn more →Credentials & frameworks the work is grounded in
Experienced, hands-on attention on every engagement, from first scope to final report.
Production code and adversarial research together, a rare combination that LLM work demands.
Every engagement ends in a clear, defensible written report, built on CISSP-grade documentation discipline.
Defined scope, fixed timeline, fixed or capped fee. No open-ended retainers or staff-augmentation arrangements.
Public writeups are proof of work, the same methodology that goes into client engagements.
A 576-trial controlled study across reasoning and non-reasoning open-weight models. Chain-of-thought reasoning lowers conventional injection rates, but attacks that target the reasoning step itself sidestep the defense, and reasoning costs roughly 15 to 20 times more tokens per call. Reasoning belongs alongside input and output controls, not instead of them.
A 1,280-trial study of indirect prompt injection susceptibility across Llama 3.1 8B, Mistral 7B, Qwen 2.5 7B, and Qwen 2.5 Coder 7B. Key finding: temperature reduction is not a reliable defense; output-format constraint achieved 0 of 40 injection on the most-susceptible model tested.
For prospective engagements in any practice area, send a note below or schedule a 30-minute scoping call.