From experimental agents to production ML pipelines.
Open Source
Live code and contributions available for public review.
A Next.js web application designed for industrial operational logging, specifically tracking material inventory and system runtime hours. The app features a secure Auth Gate for authorized users, dynamic forms built with Shadcn UI, and API routes for persisting data to an Azure SQL backend.
A Python-based integration demo for the Waste Management (WM) API, focusing on work order automation and tonnage reporting. Features include automated CSV-to-JSON ETL pipelines using Pandas, secure authentication via Base64 X-Auth headers, and a production-ready Azure Function App implementation for serverless blob-triggered processing.
A GCP-native RAG service built with FastAPI and Gemini for analyzing scientific papers. Automatically expands research collections from arXiv URLs, performs multi-document retrieval via Vertex AI Vector Search, and generates grounded, cited answers using Gemini 1.5.
PaperRec: arXiv-powered paper recommender. Harvests cs.AI/cs.LG/cs.CL/cs.RO/cs.CV, embeds abstracts (Gemini), indexes in Vertex AI Vector Search, and serves a /search API on Cloud Run.
Static portfolio site for Applied AI Engineer Evan R. Parra, built with React, Tailwind CSS, and Firebase Hosting. Features AI/BI consulting, prompt engineering for teens, and project showcases like ProfitScout and MaculaCutis.
A Next.js 15 financial analytics platform ('GammaRips') powered by MCP-based AI agents for real-time, institutional-grade stock picking. Features a frictionless 'Growth Mode' onboarding strategy and a pure Model Context Protocol architecture for dynamic market analysis.
An end-to-end, serverless AI platform built on Google Cloud that automatically ingests and analyzes financial data to generate actionable trading signals for the Russell 1000. This repository contains the full data pipeline that powers the ProfitScout application.
ProfitScout identifies explosive High Gamma options trading setups using XGBoost and technical analysis on Vertex AI. It targets volatility velocity (>0.5 ATR moves) via a fully automated MLOps pipeline with daily inference and weekly retraining.
AI-powered financial intelligence server built on the Model Context Protocol. Integrates GCP BigQuery and LLMs to automate high-gamma options strategies, market structure analysis, and technical reporting for actionable trading insights.
UrbanNexus is a governed multi-agent reasoning system designed to bridge the gap between autonomous AI capabilities and strict public sector compliance requirements. Acting as a "trust layer" for smart city infrastructure, it orchestrates a panel of specialist agents (Public Safety, Privacy Counsel, OT Security) to evaluate high-stakes deployment.
A cloud-native content generation pipeline using Google Vertex AI (Gemini 3 Flash) and a Map-Reduce architecture to automate the creation of large-scale literary analysis and multi-language artifacts.
AI-Powered Alcohol Label Verifier: A full-stack app that automates TTB label compliance checks using Google Vertex AI (Gemini 2.0 Flash) and FastAPI, deployed on Cloud Run.