1. Build end-to-end AI prototypes for government and urban use cases using LLMs, multimodal AI, RAG pipelines, agentic workflows, OCR systems, document intelligence, and VLMs. 2. Handle and process real-world datasets including PDFs, scanned documents, maps, street imagery, GPS traces, videos, spreadsheets, and public datasets. 3. Research operational domains deeply and identify bottlenecks using AI-assisted approaches. 4. Translate workflows into practical AI-driven systems and evaluate technical feasibility. 5. Experiment with emerging AI tools, frameworks, and multimodal pipelines. 6. Integrate AI systems using Claude API, LangChain, LlamaIndex, vector databases, and open-source models. 7. Develop local inference setups and scalable AI workflows. 8. Build geospatial and computer vision-based AI systems for infrastructure analysis and GIS workflows. 9. Analyze street imagery, visual scoring systems, and city-scale operational observations. 10. Ship prototypes rapidly and validate solutions through stakeholder feedback. 11. Improve systems iteratively based on testing, experimentation, and operational learnings. 12. Prepare architecture documentation, workflow notes, API documentation, deployment guides, and handoff materials.