Building traceable AI workflows for research and scientific communication.
Data Science & Big Data Technology · Yunnan University
| Project | What it does | Focus |
|---|---|---|
| research-slides | Turns papers, notes, and data into evidence-grounded research presentations. | Codex Agent Skills HTML |
| engineering-figure-gpt | Creates publication-oriented engineering figures while preserving quantitative truth. | GPT Image Python Figure QA |
| research-podcast | Converts research material into source-traceable technical audio. | Python TTS Research AI |
- Traceable by default - important claims stay connected to their sources.
- Workflow over one-shot prompting - useful patterns become reusable systems.
- Validation before polish - outputs must survive factual and structural checks.
- Tested on real work - papers, data, and technical tasks expose what toy demos miss.
AI Agents · Codex Skills · Research Automation · Multimodal AI · Scientific Computing · Data Science
Build the workflow. Test it on real work. Keep the parts that earn their complexity.

