The Literary Finder
How can specialized agents work together on a research task?
Three specialist agents research an author’s context, map their books and connect their literary legacy into a reading guide.
- Python
- LangGraph
- Gradio
Public work
Ideas, tested.
Knowledge, shared.
I use public projects to investigate AI systems, build useful tools and make technical ideas easier to understand.
01 / Explore
Questions worth building.
Prototypes and experiments in retrieval, agent coordination and cloud integration.
The Librarian
How do you ground a conversation in a specific body of knowledge?
A conversational guide to Borges that retrieves passages from a persistent document collection, using local embeddings and a language model.
- Python
- ChromaDB
- Sentence Transformers
Serverless RAG
What changes when an AI agent meets cloud constraints?
Related implementations explore how to expose a document-retrieval agent through an API on AWS and Azure.
- LlamaIndex
- AWS Lambda
- Azure Functions
FinplifAI
A chat prototype exploring specialized assistants, document tools and retrieval for Spanish accounting and finance questions.
- Next.js
- AI SDK
- TypeScript
Archived prototype, presented as an exploration of system design.
02 / Build
Tools for practical problems.
Small, focused tools for operational checks and data preparation.
Service Health Monitor
How can services report a problem before someone has to check?
Concurrent service checks with Slack alerts, an AWS Lambda entry point and Terraform configuration.
- Python
- asyncio
- Terraform
Blob Metadata Extractor
Extract file metadata in parallel and export a CSV compatible with Rails Active Storage, including content types and image dimensions.
- Python
- pandas
- Active Storage
Designed for migration preparation; results are collected in memory before CSV export.
03 / Teach
Built to teach.
Public repositories
The code, in context.
52Public repositoriesExcluding 26 forks
6Primary languagesAcross non-fork repositories
2019Earliest public repositoryIn the current non-fork inventory
GitHub snapshot · · What these numbers cover ↗
Languages across repositories
One primary language per non-fork repository, as reported by GitHub.
How to read this snapshot
78 public repositories in total: 52 non-forks and 26 forks. The non-fork count includes experiments, teaching resources, exercises and 10 archived repositories.
The language breakdown excludes 3 repositories classified as Jupyter Notebook and 3 without a primary language. It describes repository distribution, not proficiency or time spent coding.
The date marks a saved public snapshot, not a live activity feed. Private work, private contributions and client outcomes are outside its scope.
See the repositories on GitHub