profile sourced from GitHub
Scouting report
Puerto Rican junior AI dev building offline-first RAG copilots in Python
assessed from open-source footprint
Danny is a new (six-month-old account) Puerto Rican developer focused squarely on applied AI/ML in Python and TypeScript, with nothing abandoned yet. His flagship Nova RAG is a safety-focused, offline-first RAG copilot with hallucination-mitigation and eval tooling; Project Lumina pairs a physics sim with a Gemini narration layer. Early-stage and lightly starred, but the focus on local LLMs, retrieval and safety is clear and deliberate.
Authorship & open source
What they build
Industry experience
- Data, ML & AI
- Media & Streaming
Signal breakdown
2
top repo 2
8
0% forks
2
0.5 yr
0
Active
0% stale
Strengths
- Verified author — wrote 100% of commits on drosadocastro-bit
- 18 merged pull requests
- 479 commits in the past year
- Original builder — 8 of their own repositories
- Consistently active, low abandonment
- Data / ML focus with Backend
- Domain experience in Data, ML & AI & Media & Streaming
About
Skills
- Python
- TypeScript
- Ai
- Local Llm
- Offline
- Rag
- Retrieval Augmented Generation
- Safety Critical
Featured work
Nova Rag Public
Nova Intelligent Copilot A safety-focused, offline-first RAG copilot with hallucination mitigation and evaluation tooling.
- Python
- Ai
- Local Llm
- Offline
- Rag
by Danny Rosado Castro
Project Lumina
Lumina: A well-designed physics simulation with a Gemini reflection layer that generates compelling narration, an audio coupling mechanism that genui…
- TypeScript
by Danny Rosado Castro
Manatuabon
Manatuabon is a local-first, human-governed astrophysics reasoning workspace with optional Claude cloud escalation for deep review. It combines struc…
- Python
by Danny Rosado Castro
Julia
Julia is an AI-powered crop caretaker system designed for home gardeners in Puerto Rico.
- Python
by Danny Rosado Castro
Agent K
Agent K is a standalone, offline-first integrity scoring tool for AI agents.
- Python
by Danny Rosado Castro
Cibuco Boriken
CFAR-inspired adaptive thresholding from radar signal processing, adapted for bioacoustic species detection.
- Python
by Danny Rosado Castro