LLM Engineering & Security

LLM application security & EU AI Act adversarial testing.

Custom LLM builds, on-premises deployment, and adversarial security testing for organizations putting AI into production, including documented EU AI Act adversarial-testing evidence for the August 2026 Article 15 deadline. Project-based engagements with written deliverables.

1856Prompt-Injection Trials Published
OWASPLLM Top 10 · MITRE ATLAS
On-PremSelf-Hosted Deployment

Most AI security people can't ship code. Most LLM developers don't think about security architecture.

Article 15
Robustness & security
Reproducible
Payloads & harness
EU AI Act · Article 15

EU AI Act adversarial-testing evidence, before the August 2026 deadline

For US companies with EU customer exposure, Article 15 of the EU AI Act sets obligations for accuracy, robustness, and cybersecurity. Demonstrating compliance means producing documented adversarial-testing evidence, not a checkbox. Secure Data Consortium delivers that evidence as a structured, defensible written package in the format regulators expect.

Robustness & security testing

Adversarial probing for prompt injection, jailbreaks, and manipulation, mapped to OWASP LLM Top 10 and MITRE ATLAS.

Evidence in the right format

Methodology, reproducible payloads, results, and remediation, documented to support an Article 15 conformity narrative.

Re-runnable for future versions

A test harness your team keeps, so the evidence can be regenerated as models and features change.

Discuss an Evidence Package
What This Practice Does

Three sides of the same expertise: build it, secure it, prove it

Over a decade of enterprise software architecture combined with active adversarial security research, so the same practice can develop your LLM features, test them like an attacker, and document the evidence regulators ask for.

Development & Engineering

Putting LLMs into production

Application development & integration

Custom LLM-backed applications and integration into your existing systems. The deliverable is working, documented code, not slides.

RAG pipelines for your documents

Retrieval pipelines built against your own corpus (case files, clinical guidelines, policy documents, internal knowledge bases) so answers are grounded in your content.

On-premises & self-hosted deployment

For organizations that can't send data to a third-party API. Secure Data Consortium installs, configures, and hardens an open-weight model on your own infrastructure, air-gapped or network-isolated where required, and hands back a running system your team owns and controls, with documentation to operate it.

Security & Adversarial Testing

Finding the failure modes before someone else does

Security review & adversarial testing

A review of how untrusted content reaches your model and where it can be turned against you, followed by hands-on adversarial testing: prompt injection, tool-use hijacking, system-prompt extraction. You get a written report with reproducible findings, remediation priorities, and results mapped to the OWASP LLM Top 10 and MITRE ATLAS.

Reusable evaluation & red-team harness

Test infrastructure your team keeps and re-runs against every future model version and feature change, so security testing becomes part of your release process, not a one-time report.

Compliance & Evidence

Turning testing into documentation that holds up

EU AI Act adversarial-testing evidence

For US companies with EU customer exposure preparing for the August 2026 obligations under Article 15: documented adversarial-testing evidence for robustness and cybersecurity, in the format regulators expect, and re-runnable as your models change.

AI risk & governance documentation

Security and risk documentation aligned to recognized frameworks (NIST AI RMF, ISO/IEC 42001) that demonstrates a defensible testing and risk-management process to auditors, partners, and customers.

Why This Practice

Builder and breaker, in one practitioner

Builder & Architect

Enterprise architecture and application development at Bank of America, Citigroup, Verizon, Walmart, and others.

Active Adversarial Research

Cross-model prompt injection studies on locally-hosted LLMs (Ollama). Published, reproducible methodology on GitHub.

Shipping Portfolio

Production algorithmic trading system on the Coinbase Advanced Trade API; custom Python and Go security tooling.

Registered Researcher

Registered security researcher on the Coinbase HackerOne program. Previously a Customer Success Engineer at Chainalysis.

From the Lab

Independent research, published in the open

Reasoning Defense · 576 trials

Does reasoning make LLMs safer against prompt injection? Testing Qwen 3, DeepSeek-R1, and Gemma 4

A 576-trial controlled study across four reasoning conditions (Qwen 3 8B thinking off and on, DeepSeek-R1 8B, and Gemma 4 e4b), spanning eleven attack techniques, four application scenarios, and three trials each. Chain-of-thought reasoning reduces conventional prompt-injection susceptibility (Qwen 3: 64% to 54% genuine injection between non-reasoning and reasoning), but payloads that target the reasoning step itself sidestep the defense, and reasoning costs roughly 15 to 20 times more tokens per call. The conclusion: reasoning belongs alongside input and output controls, not instead of them.

Prompt Injection · 1,280 trials

Temperature is not a defense: indirect prompt injection across four open-weight LLMs

A 1,280-trial study characterizing indirect prompt injection susceptibility across Llama 3.1 8B, Mistral 7B, Qwen 2.5 7B, and Qwen 2.5 Coder 7B at two production-realistic temperatures. Temperature reduction is not a reliable defense; output-format constraint achieved 0 of 40 injection on the most-susceptible model tested.

How Engagements Work

Built to deliver, without the embedded-contractor failure modes

Project-shaped, not staff augmentationEvery engagement has a defined scope, a written deliverable, and a fixed timeline. No "embedded contractor" or "extension of your team" arrangements.
Defined response windowsEngagement letters specify response windows, typically two business days, not availability. Clear expectations, no on-call overhead.
Fixed-scope or capped-hours pricingMost engagements are fixed-fee per deliverable. Hourly work is capped per engagement; no open-ended hourly retainers.
Remote-first deliveryEngagements run via VPN or jump box for environments that require it. On-site days are bounded and scheduled at engagement start.
Start a Conversation

The fastest path is a 30-minute scoping call

Tell us what you're working on, or schedule directly.

Schedule a Scoping Call