Multimodal agent
Maintenance-Eye
Multimodal maintenance copilot with live audio, camera input, backend tools, deployment, and tests.
Google ADK, Gemini Live API, FastAPI, Firestore, Cloud Run
Applied AI Engineer
Applied AI Engineer building multimodal agents and LLM systems for operational workflows.
I build production-minded AI systems where models are one part of a working product: real-time interfaces, backend tools, retrieval, confirmations, deployment, and tests.
Maintenance-Eye is a public case study for a real-time maintenance copilot using Google ADK, Gemini Live API, FastAPI, Firestore, and Cloud Run. GovIntel demonstrates federal-procurement RAG with ingestion, hybrid retrieval, SQL analytics, citation validation, FastAPI, and Streamlit.
- Flagship
- Maintenance-Eye
- RAG System
- GovIntel
- Domain Context
- Transit maintenance workflows
Featured Project
Open case study →
Google Gemini Live Agent Challenge project
Maintenance-Eye
A real-time maintenance copilot that streams camera and microphone input from a phone to a FastAPI backend, orchestrates Google ADK tools, and uses Gemini Live API for native multimodal interaction.
- Tool workflows for asset lookup, knowledge retrieval, safety guidance, and work-order actions.
- Human-in-the-loop confirmation for critical actions.
- Cloud Run, Firestore, Docker, Terraform, GitHub Actions, and multi-layer tests.
Case Studies
All projects →RAG system
GovIntel
Federal procurement intelligence system with USAspending ingestion, hybrid retrieval, SQL analytics, citation validation, API, and UI.
Python, FastAPI, Streamlit, PostgreSQL, Chroma, BM25
Published research
Autism Screening ML Research
Published ML research on an autism-screening data dashboard and SVM/CNN classification literature.
Python, Tableau, SVM/CNN methods