Book
Work

Built, deployed, running.

These are real systems — each with a public GitHub repo you can read. Built and shipped on Google Cloud Platform. Full case studies and additional projects at danielflugger.com.

Geospatial ML · Open Source

GeoAsset Verify Pipeline

Automated Property Verification via Geospatial Intelligence

Production ML pipeline that cross-references property records, parcel boundaries, and document-claimed attributes against live geospatial data. Flags discrepancies between recorded and actual property characteristics using PostGIS spatial analysis and Vertex AI anomaly detection. Built for real estate due diligence, insurance underwriting, and municipal assessment workflows.

View on GitHub →

GeoAsset Verify architecture diagram

What It Does

Automated parcel-level property verification — replaces manual record-to-reality checks with ML-driven spatial analysis

GCP Services

Vertex AI (model training + serving), BigQuery (analytical pipeline), Cloud Run (API), Google Maps Platform (spatial data)

Stack

PostGIS Vertex AI BigQuery Python Cloud Run
RAG / LLM · Open Source

RAG Doc Processor

Grounded Document Intelligence for Engineering Workflows

Production RAG system that ingests multi-source engineering documents — RFPs, proposals, inspection reports — and generates structured outputs using grounded LLM retrieval via Vertex AI. Designed for domains where document accuracy has financial and legal consequences. Significantly reduces manual review cycles while maintaining citation traceability.

View on GitHub →

RAG Doc Processor architecture diagram

What It Does

Processes engineering documents into structured, grounded outputs with full citation tracking — replacing manual review workflows

GCP Services

Vertex AI (embeddings + generation), Vector Search (retrieval), Cloud Run (serving), Cloud Storage (document ingestion)

Stack

Vertex AI FastAPI Pydantic Vector Search Python
LLM Orchestration · Open Source

Vendor Mesh Intelligence

AI-Driven Procurement Intelligence

LLM-powered procurement intelligence that aggregates several hundred vendor sources, applies Gemini-powered semantic scoring to match vendors against project briefs, and generates structured comparison reports. Built to replace weeks of manual vendor research in construction, manufacturing, and enterprise procurement.

View on GitHub →

Vendor Mesh Intelligence architecture diagram

What It Does

Automates vendor discovery and evaluation — semantic matching across hundreds of sources, structured comparison output

GCP Services

Gemini API (semantic scoring), BigQuery (vendor data warehouse), Cloud Run (API + frontend), Pub/Sub (ingestion pipeline)

Stack

Gemini API BigQuery React Pub/Sub Python
Voice Biometrics · ML Pipeline

Vera VB

Voice Biometric Identity Platform

Zero-knowledge privacy architecture for voice-based identity verification. Five-layer verification stack processing acoustic features through Resemblyzer, Wav2Vec2, Parselmouth, and librosa for multi-dimensional speaker analysis. Client-side encryption via Web Crypto API, blockchain timestamping via OriginStamp. No raw audio stored server-side.

Vera VB architecture diagram

What It Does

Voice-based identity verification with zero-knowledge privacy — five-layer ML stack, client-side encryption, no raw audio on server

GCP Services

Cloud Run, Cloud Functions, Firebase Auth, Firestore, BigQuery — full GCP-native deployment

Stack

Cloud Run Firebase Auth Firestore BigQuery Next.js Python
Reference Architectures

Integrated platform patterns.

Architecture case studies built on the production stack — ADK, A2A, MCP, Claude, and Gemini on Google Cloud. These represent the system design patterns I deploy for clients.

Platform Design · Manufacturing

Supply Chain Command Center

Multi-System Orchestration for Manufacturing Operations

Integrated operations platform connecting ERP, warehouse management, and logistics systems via MCP tool servers. Claude handles purchase order analysis, vendor communication, and exception reasoning. Gemini handles real-time classification and routing of shipment data. ADK orchestrates specialized sub-agents for procurement, inventory, and fulfillment — coordinating via A2A protocol across organizational boundaries with supplier systems.

Supply Chain Command Center dashboard

What It Does

Unifies fragmented supply chain systems into a single orchestration layer — procurement intelligence, inventory optimization, logistics coordination with supplier interop

GCP Services

Vertex AI (Claude + Gemini), Cloud Run (services), Pub/Sub (event bus), BigQuery (analytics), Firestore (state), Cloud Monitoring

Stack

ADK A2A MCP Claude on Vertex AI Gemini Cloud Run Terraform
Platform Design · Regulated Industries

Compliance Intelligence Platform

Automated Regulatory Mapping & Gap Analysis

Document intelligence platform for regulated industries — ingests regulatory frameworks, internal policies, and audit documentation via structured retrieval. Claude performs deep reasoning over compliance requirements, mapping regulatory language to internal controls and identifying gaps. Generates structured compliance reports with citation traceability. Model Armor enforces data governance at every interaction boundary.

Compliance Intelligence Platform dashboard

What It Does

Transforms regulatory compliance from manual review into automated gap analysis — maps requirements to controls, flags exposures, generates audit-ready reports

GCP Services

Vertex AI (Claude + embeddings), Vector Search, Document AI, Cloud Run, Model Armor, BigQuery (compliance warehouse)

Stack

Claude on Vertex AI Document AI Vector Search Model Armor FastAPI Pydantic
Platform Design · Enterprise

Enterprise Knowledge Fabric

Unified Retrieval Across Organizational Knowledge

Internal knowledge platform connecting fragmented data sources — Google Workspace, Confluence, Slack, JIRA, internal databases — through MCP tool servers into a unified retrieval layer. Claude synthesizes multi-source answers with full attribution. Gemini Flash handles high-volume classification and routing. ADK manages specialized retrieval agents, each connected to different organizational knowledge domains.

Enterprise Knowledge Fabric architecture

What It Does

Turns scattered organizational knowledge into a searchable, intelligent system — grounded answers with sources from across every connected tool

GCP Services

Vertex AI (Claude + Gemini Flash), Vector Search, AlloyDB (structured data), Cloud Run, Agent Engine, IAM

Stack

ADK MCP Claude on Vertex AI Gemini Flash AlloyDB Agent Engine
Platform Design · Operations

Field Operations Hub

Intelligent Coordination for Distributed Teams

Operations platform for companies with distributed field teams — integrates scheduling, asset tracking, inspection reporting, and work order management. Field technicians submit reports through a mobile interface; Claude processes inspection data into structured findings with severity classification. Gemini handles image analysis of on-site conditions. Event-driven architecture on Eventarc and Cloud Run generates daily summaries, flags anomalies, and coordinates resource allocation across regions.

Field Operations Hub dashboard

What It Does

Real-time visibility across distributed field teams — inspection intelligence, automated scheduling, anomaly detection, resource optimization

GCP Services

Vertex AI (Claude + Gemini Vision), Cloud Run, Eventarc, Firestore (real-time state), Maps Platform, Cloud Monitoring

Stack

ADK Eventarc Claude on Vertex AI Gemini Vision Maps Platform Firebase
Current Engagements

I'm currently engaged on two private projects involving regulatory document analysis and predictive maintenance infrastructure — both on GCP, both under NDA. Happy to discuss the type of work in conversation.

Your Project

Something to build?

If you're working on a GCP project and need an engineer who ships, let's talk.

Book a Project Review →