• AI Strategy Consultant

Designed and implemented an AI-powered operations environment for a business owner managing multiple businesses across property management, agriculture, wellness, and finance. Connected Claude agents, Google Sheets, email, business applications, and custom dashboards to automate data maintenance, verification, financial monitoring, and reporting. Key Systems Built Scheduled AI Agents: Implemented 8 scheduled tasks for property management, agriculture, wellness operations, billing, and urgent financial alerts. Human-in-the-Loop Workflows: Built separate verifier and updater agents. Verifiers audit spreadsheets, emails, financial records, and documents, then email proposed changes and discrepancies for human review. Updater agents process approvals and corrections by email and apply authorized changes. Operational Dashboards: Developed 5 Claude-powered dashboards connected to business data, covering property management, agriculture, wellness, and financial reporting. Data Integrity: Implemented source-backed updates, deduplication, business-specific data isolation, change logs, and evidence tracking. Unsupported or conflicting information is flagged, never guessed. Billing Automation: Built a fortnightly agent monitoring 5 email inboxes and maintaining 71 recurring bills across 6 entities, including payment statuses, cancellations, and deadlines. Live Data & Reporting: Implemented dashboard refresh mechanisms to retrieve updated spreadsheet data and synchronize operational changes. Business Integrations: Configured 4 MCP integrations using Maton, NuExtract, Agriculture platform and Pinecone, connecting 40+ applications, including QuickBooks, Google Workspace, Twilio, and WhatsApp. Additional Automation: Built email-to-SMS notifications, structured research reporting, and local database extraction and cleanup workflows. Cost Optimization & Business Impact Audited software, cloud infrastructure, AI services, and subscriptions to eliminate unnecessary expenses and optimize plans. ~$35,685/year in potential savings after a pending cancellation. 8 scheduled tasks, 5 dashboards, 40+ application integrations, and 3 MCP integrations. 71+ recurring bills consolidated into a centralized register. My Role Forward Deployed AI Engineer & Technical Lead Owned workflow architecture, Claude agent development, scheduled automation, MCP integrations, dashboards, data verification, approval workflows, deployment, and cost optimization. Technology Stack Claude, Claude Code, MCP, Maton, NuExtract, Pinecone, Google Sheets, Claude Artifacts, QuickBooks, Twilio, WhatsApp, Google Workspace, Email Automation, Scheduled Agents, Human-in-the-Loop Systems, Data Validation, Audit Logging, Context Engineering.

• AI Developer
• AI Engineer
• AI Ethics Consultant
• +2

Built a multi-agent AI platform connecting 40+ systems across six businesses, delivering 21 automated workflows, custom MCP infrastructure, and approximately $35K in annual savings. Overview A production multi-agent AI platform I designed and built for a founder managing six businesses. It connects 40+ business systems through 21 automated workflows and three custom Model Context Protocol (MCP) servers, providing a centralized AI-powered operational command center. The Problem Business operations were fragmented across disconnected applications, inboxes, databases, and documents. Manual financial workflows, unreliable knowledge retrieval, and limited operational visibility created inefficiencies across the businesses. The Solution Built 21 automated workflows with specialized, tool-calling agents connecting 40+ business systems, including QuickBooks, CRM platforms, and Google Workspace. Developed three custom MCP servers with structured agent-to-system communication, three autonomy tiers, per-business credential isolation, OAuth integrations, and human approval gates. Developed agent execution infrastructure with reusable skills, context memories, multi-agent configurations, and context optimization. Rebuilt the production retrieval pipeline, improving retrieval coverage from 0.3% to over 90% with source citations. Implemented durable agent execution, checkpointing, retries, monitoring, and offline evaluation. Identified and fixed five silent failures. Reduced a $1,000+/week LLM cost overrun through context pruning and intelligent Claude model routing. Business Impact & Results ~$35K in total annual savings. 15 hours/week of manual work eliminated through invoice and payment exception automation. ~60% faster incident resolution. 0.3% → 90%+ retrieval coverage. $1,000+/week in LLM cost overruns eliminated. 5 production failures identified and fixed. My Role Forward Deployed AI Engineer & Technical Lead Owned end-to-end delivery, from client discovery and AI architecture to agent development, backend engineering, MCP integrations, testing, deployment, and production optimization. Built using Claude Code and specification-driven, AI-assisted software development. Technology Stack & Engineering Expertise Agentic AI: Multi-Agent Systems, LangGraph, Claude, OpenAI, Agent Orchestration, Tool Calling, Model Routing. Context & Retrieval: Context Engineering, Agent Skills, Context Memories, RAG, Semantic Search, pgvector, Pinecone, NuMind. MCP & Integrations: Custom MCP Servers, Multi-Tenant Architecture, OAuth, QuickBooks, HubSpot, Salesforce, Google Workspace, n8n. Backend & Cloud: Python, FastAPI, PostgreSQL, AWS, Event-Driven Architecture, Durable Execution. Production AI: LangSmith, LLM Observability, Evaluation, Regression Testing, Agent Monitoring, Cost Optimization, Claude Code.

• AI Developer
• AI Engineer
• AI Ethics Consultant
• +2