Operators, not headcount.
Azcuna Dynamics is a small, technical team focused on a hard problem: making foreign-language material usable where the cloud is not available and the data cannot leave. The size is deliberate. The systems we build have to be understood end to end by the people who answer for them in an accreditation review, and that is easier with a team that fits in one room.
What a small team cannot fake is a track record, so this page leads with the record: the patents, the research background, and the working software.
Founded Azcuna Dynamics to bring offline, sovereign language AI to defense and intelligence teams operating beyond the reach of the cloud. Named inventor on two granted U.S. patents and a former data-science researcher at the Harvard Kennedy School. Works in English and Spanish.
- Patents
- Two granted U.S. patents: a quantum-enhanced decryption process for RSA and AES, and an adaptive feedback system for ballistic accuracy
- Research
- Data science, Harvard Kennedy School
- Built
- BABEL offline multilingual core; ELLMENT prototype (recorded session on this site)
- Languages
- English, Spanish
Core engineering stays in-house
Models, retrieval, the inference runtime, and the security posture are built and maintained by the founding team. That is the part of the system an evaluator is accrediting, and it does not get subcontracted.
Domain experts by program
Each program brings in the operators and subject-matter experts it needs: Naval Flight Officers and E-2D community SMEs for ELLMENT, linguists and analysts for BABEL evaluations. They shape the corpus, the question sets, and the interface. They are named in the proposal, not hidden in an org chart.
Advisors under agreement
We list advisors only after a signed advisory agreement. Announced names on a website are cheap; accountable ones are not.
We are assembling an advisory board of senior defense, intelligence, and language professionals. Names are published only under signed agreements. Announcements to follow.
If you are an operator, analyst, or linguist who has fought a language problem in a place with no network, we want to hear how you did it. If you are a program office evaluating offline AI, we will demonstrate on a laptop with the radio off. If you are an engineer who wants to work on retrieval, translation, or inference at the edge, write to us.