Challenge
Process large mixed-media sources into trustworthy answers without losing timestamps, speaker context or links back to original evidence.
Featured projects
Video & document intelligence
AI RAG Assistant
An enterprise research assistant that turns video, audio and documents into searchable, cited knowledge.
The project
Turning a complex workflow into a dependable product.
Approach
Create a modular ingestion pipeline for transcription, diarization, chunking, metadata, embeddings and retrieval, with citations preserved throughout the answer flow.
Contribution
- Designed the ingestion and retrieval architecture
- Built transcript, passage and embedding workflows
- Implemented filtered search and source citations
Outcome
A swappable Rails-based AI platform that can evolve across models, storage and retrieval strategies without rewriting the product.
Contact
Let's build something great.
Open to senior engineering opportunities and meaningful collaborations. Tell me what you're working on and I'll get back to you.
rorens05@gmail.com