projects

Projects

A selection of things I have built: applied AI, data analysis and tooling for agent-driven development.

AIProfessional case study · banking · 2026

AI agents for the development lifecycle

2xproductivity in the 3 teams using it

An agentic spec-driven development (SDD) system for GitHub Copilot, with custom agents and skills that support development teams from discovery to QA.

  • Agents for discovery, development and QA
  • Jira, Figma and internal-system integrations through MCP
  • Mandatory human review at every stage
GitHub CopilotMCPSDDJiraFigma
AIPersonal project · 2026

Ask about my background

$0cost, within the free tier

An AI assistant on this very site that answers questions about my experience, projects and publications, using RAG over the site's content and citing its sources.

  • Only answers with what is already published on the site
  • Cites the pages each fact comes from
  • Daily limits so it never leaves the free tier
RAGCloudflare WorkersWorkers AITypeScriptAstro
ToolingCommunity preset · v1.0.1 · 2026

Spec Kit · coordinated multi-repo branching

3repo detection modes

A community preset for GitHub Spec Kit that detects which repositories a feature affects and generates the tasks needed to create matching branches in all of them.

  • Supports independent repos and git submodules
  • Extends /speckit.plan and /speckit.tasks
  • Works with the Brownfield extension
Spec KitSDDAgentsGit
DataMaster's thesis · 2026

Canton-level crime risk index for Ecuador

−46%RMSE vs. ARIMA

A composite crime risk index covering every canton in Ecuador, combined with KMeans segmentation and a national forecast benchmarked across six time-series models.

  • Open data on homicides, arrests and seized weapons
  • 6 canton profiles by level and type of risk
  • The ensemble model reduces RMSE by ~46% compared to ARIMA
PythonTime seriesLSTMXGBoostPCAKMeans
AI2026

Chat with scientific papers (RAG)

100%local, no external APIs

A local RAG app for chatting with academic papers, featuring section-aware chunking, MemoRAG and re-ranking. It runs entirely on open-source models through Ollama.

  • LangGraph pipeline with standard RAG and MemoRAG modes
  • Two-stage retrieval with a cross-encoder
  • Page-level citations in English or Spanish
LangGraphRAGOllamaChromaDBStreamlit
DataData Visualization course · 2026

AI research trends (OpenAlex)

70+years of AI research

An interactive visual analysis of more than 70 years of AI research using OpenAlex data: growth, emerging topics, geography, collaboration and open access.

  • 7 research questions, from 1950 to today
  • Interactive treemaps and maps
  • Published as a Quarto site
PythonAltairPlotlyQuartoOpenAlex