Building AI infrastructure for languages, knowledge systems, and institutions that cannot afford to be treated as ordinary data.
MoniGarr is an independent research and engineering practice founded by Monica Peters (“MoniGarr”), working at the intersection of artificial intelligence, language, computational linguistics, software architecture, cultural knowledge, sovereignty, and long-term institutional trust.
The practice develops research, architectures, infrastructure, and standards for AI systems operating in environments where language and cultural context are not peripheral concerns—but foundational requirements.
MoniGarr is the founder of MohawkLanguage.ca and the Language Intelligence initiative.
The work begins with Kanien’kéha and extends to a broader question facing the AI field:
What should AI architecture look like when human language is treated as a living knowledge system rather than simply another category of training data?
If you are researching, funding, governing, building, or deploying consequential AI systems involving language, culture, knowledge, or sovereignty, this work is for you.
Language Intelligence
A diA research and engineering discipline for building AI around the structure, knowledge, sovereignty, and lived reality of human languages.
Through MohawkLanguage.ca, MoniGarr develops Language Intelligence infrastructure for Kanien’kéha dialects, including:
Language datasets · Computational linguistics · Speech technology · ASR · TTS · Pronunciation systems · AI evaluation · Language models · Educational systems · Sovereign language infrastructure
The objective is to make existing AI models support polysynthetic language dialects and understand what must change when the language itself has different computational structures, different knowledge relationships, different histories, different governance requirements, and different communities of authority.
Language Intelligence asks questions conventional AI architecture often leaves unanswered.
- How should AI represent polysynthetic morphology?
- How should speech systems be evaluated when pronunciation and dialect variation matter?
- Who owns the resulting linguistic data?
- How should culturally governed knowledge enter or remain outside an AI system?
- How do communities retain meaningful authority over language infrastructure?
- How should models be evaluated when conventional multilingual benchmarks fail to measure what matters?
And ultimately:
What does sovereign, culturally responsible AI look like at the architectural level?
Research → Engineering → Evaluation → Standards → Education → Infrastructure
The AI industry has a language problem.
Modern AI systems are extraordinarily capable. Their assumptions are not universal. Adding a language to a multilingual model does not necessarily mean the language has been adequately represented.
Polysynthetic languages, low-resource languages, Indigenous knowledge systems, oral traditions, culturally governed datasets, community-owned infrastructure, and radically different morphological structures introduce architectural problems that conventional AI systems were not designed to solve.
Language Intelligence begins there. It examines the entire intelligence system:
Language → Data → Models → Speech → Evaluation → Governance → Applications → Institutions
The question is not only whether an AI system can produce an answer in a language.
The question is whether the system can interact with that language accurately, responsibly, transparently, and under the authority of the people and institutions to whom that language belongs.
LANGUAGE INTELLIGENCE STACK
Linguistics
Morphology · Syntax · Lexicon · Semantics · Pragmatics
Speech
ASR · TTS · Pronunciation · Speech Corpora · Voice Technology
Data
Corpora · Metadata · Provenance · Annotation · Data Sovereignty
Intelligence
LLMs · SLMs · Agents · Retrieval · Reasoning · Evaluation
Governance
Ownership · Consent · Cultural Safety · Access · Accountability
Infrastructure
APIs · Tools · Research Platforms · Educational Systems · Community Infrastructure
Core Research Domains
Indigenous Language AI
Architecture and engineering for Indigenous, polysynthetic, and low-resource languages, with attention to linguistic structure, cultural ownership, community authority, and data sovereignty.
Language-Centered AI Evaluation
Research into how AI systems should be evaluated when conventional multilingual benchmarks fail to capture linguistic structure, dialect variation, cultural context, or community-defined measures of success.
Sovereign AI & Knowledge Infrastructure
Architectures for AI and knowledge systems in which ownership, governance, provenance, access, and institutional authority are first-class technical requirements.
AI Governance & Cultural Risk
Technical and architectural analysis of AI systems where cultural harm, inappropriate representation, misuse, or erosion of institutional trust creates material risk.
Research & Infrastructure Modernization
Transformation of fragmented, legacy, or incoherent research and AI environments into durable, reproducible, interoperable infrastructure.
High-Consequence AI Systems
System-level architecture and stewardship for deployments where technical failure can become cultural, social, institutional, or reputational failure.
From Language to Infrastructure
MoniGarr works across the full lifecycle of an intelligence system:
Language → Data → Models → Evaluation → Governance → Applications → Institutions
The objective is not simply to build another model or another application.
It is to build systems that can remain:
Useful. Understandable. Evaluatable. Governable. Sovereign. Trustworthy.
And capable of remaining meaningful long after the technology that originally produced them has changed.
Why This Work Matters
The next generation of AI will encounter a world far more linguistically, culturally, and institutionally diverse than the datasets and architectures that dominate today’s systems.
The difficult problems will increasingly exist at the boundaries:
AI × Language
AI × Culture
AI × Governance
AI × Sovereignty
AI × Institutions
AI × Long-Term Knowledge
These are not problems that can be solved by model size alone.
They require people who can move between disciplines without losing the structure of any of them.
That is the territory of MoniGarr.
Authority
MoniGarr combines more than three decades of technology and software engineering experience with ongoing work in artificial intelligence, machine learning, computational linguistics, language revitalization, digital media, research infrastructure, and language technology.
The practice is deliberately interdisciplinary.
Its work connects technical architecture with the realities of living languages, communities, institutions, and long-term stewardship.
The result is a perspective that does not fit neatly inside conventional AI, software engineering, linguistics, or cultural-sector categories.
That is intentional.
A Different Kind of Practice
MoniGarr is intentionally independent.
The practice is designed to operate between disciplines that are often separated inside conventional organizations:
AI · Linguistics · Engineering · Cultural Knowledge · Governance · Education · Infrastructure
This allows problems to be examined across the boundaries where context is often lost.
MoniGarr is particularly interested in problems that are:
- technically difficult
- interdisciplinary
- under-researched
- institutionally consequential
- culturally sensitive
- poorly served by conventional solutions
- likely to become more important over the next decade
Collaboration
MoniGarr works with a limited number of institutions, research programs, funders, and collaborators where there is strong alignment between the problem, the architecture, and the potential long-term impact.
The preferred forms of collaboration include:
Sponsored research
Institutional research appointments
Fellowships and residencies
Strategic architecture
Research partnerships
Standards and infrastructure initiatives
Selective advisory appointments
The practice prioritizes research depth, architectural authority, continuity, and long-horizon stewardship over transactional deliverables.
Precedent
MoniGarr’s work spans software architecture, artificial intelligence, language technology, language revitalization, digital media, education, research, and cultural infrastructure across North America and internationally.
Selected research, projects, collaborations, and technical work are documented throughout this site.
Additional institutional work may be shared selectively due to confidentiality and the nature of the engagements.
The Work Ahead
The next generation of AI will not be defined only by larger models.
It will also be defined by whether those systems can responsibly operate across the extraordinary diversity of human languages, knowledge systems, cultures, institutions, and forms of sovereignty that exist beyond the assumptions of today’s dominant AI infrastructure.
MoniGarr is building toward that future.
For Researchers, Institutions & Builders
If you are working on a problem involving:
Language AI · Indigenous Language Technology · Computational Linguistics · Speech AI · Low-Resource Languages · AI Evaluation · Sovereign AI · Cultural Data · Knowledge Systems · AI Governance · Research Infrastructure · Responsible AI
there may be meaningful territory to explore together.
MoniGarr does not pursue volume.
The practice is interested in the problems that matter.
Institutional Inquiry
MoniGarr works with a limited number of institutions and research partners each year.
Inquiries are evaluated for:
Intellectual alignment · Institutional fit · Research significance · Long-term potential · Meaningful impact
