I'm a sociologist and applied AI practitioner working at the intersection of the social sciences, humanities, and engineering. I study the history of sociological theory, train AI systems for data analysis, document understanding, research, and knowledge work, and build end-to-end ReAct agents. I turn these systems into usable products with reproducible Docker deployments and user-friendly Vue.js interfaces.
| Project | What it does | Focus |
|---|---|---|
| Academic Research Skills | An agent-skill toolkit for structured research, academic writing, review, revision, and publication workflows. | AI agents · Research workflows |
| Education Agent Skills | A collection of evidence-based skills for AI agents that support teaching, learning, instructional design, and education workflows. | AI in education · Agent skills |
| GigaAMGUI | A complete interface for GigaAM-v3 transcription: desktop app, CLI, REST API, web UI, speaker diarization, and Docker deployment. | Speech AI · Python · APIs |
| LaTeX → Word Formula Converter | Turns raw LaTeX math in .docx files—including LLM-generated documents—into editable native Word equations (OMML). |
Document automation · PyQt6 · Pandoc |
| MDtoWORD | A desktop application for converting Markdown documents into fully formatted Microsoft Word files, with batch processing and a PyQt6 interface. | Document automation · Python · PyQt6 |
- Theoretical sociology — the history and development of sociological theory, its concepts, schools, and intellectual traditions.
- Data analysis — turning complex social, research, and product data into structured evidence and useful insights.
- Interdisciplinary AI — training and applying AI at the intersection of social sciences, humanities, and technical disciplines.
- Agentic and ReAct systems — end-to-end agents with reasoning–action loops, tool use, memory, planning, orchestration, and systematic evaluation.
- Applied LLM systems — assistants, RAG-style workflows, model integrations, and useful automation.
- Local inference and model serving — efficient local execution with llama.cpp and high-throughput serving with vLLM.
- OCR and document vision — text and layout extraction with Tesseract, PaddleOCR, LightOnOCR, and vision-language models (VLMs).
- Speech and language technology — transcription pipelines, diarization, and interfaces that make AI models accessible.
- Document intelligence — converting AI-generated content into reliable, editable Office documents.
- Product delivery — containerized services with Docker and responsive frontends built with Vue.js.
- Xray networking — Xray-core configuration, policy-based routing, load balancing, and Observatory-driven outbound health checks.
- AI-assisted development — agent workflows, persistent project context, testing, and code-quality automation.
- Research and knowledge workflows — tools that help turn complex source material into useful, structured outputs.
Languages & platforms
Areas of practice
- Computational sociology, sociological theory, and data-driven research
- AI tools for social sciences, humanities, and interdisciplinary scholarship
- Production-grade ReAct agents and multi-agent systems
- Open-source tools for applied AI and local models
- OCR, document vision, and multimodal AI products
- Speech, transcription, and knowledge-management products
- Document-processing and academic research workflows
- Dockerized services and Vue.js applications
- Xray-core routing, balancing, and resilient proxy infrastructure
- Developer tooling that makes AI agents more dependable
I bridge sociological thinking, humanistic inquiry, and engineering practice. I care about turning capable models into tools people can actually use: clear interfaces, reproducible setup, practical documentation, and workflows that hold up outside a demo.