Benjamin – AWS, Python, LLM

Developed an AI Admission Agent for a US Mental Health Rehab Center to reduce costs and improve patient intake and communication workflows.

• Created prompts for the AI Admission Agent to handle calls, gather relevant information, and manage patient interactions;
• Implemented backend workflows using n8n and JavaScript;
• Developed AI simulations and regression datasets for pre-deployment testing;
• Debugged live calls and implemented prompt fixes.

Developed an AI Voice Agent from scratch to handle calls, collect relevant information, and efficiently dispatch technicians for a US-based property restoration company specializing in biohazard cleanup.

• Created prompts for the AI Voice Agent to handle calls, gather relevant information, and dispatch technicians via GoHighLevel;
• Implemented backend workflows with n8n and JavaScript;
• Developed AI simulations and regression datasets for pre-deployment testing;
• Debugged live calls and applied prompt optimizations to improve agent performance.

Minute of Glory: A1 Biohazard now saves over $26,000 per year and can scale client volume without any additional cost.

A real estate agent required an efficient way to target high-value prospects (agents with $4M+ in sales) and book appointments to showcase a white-labeled AI voice solution.

• Created effective prompts for the AI Agent and optimized responses to ensure natural, empathetic interactions;
• Built backend workflows in n8n and JavaScript to connect tools and automate processes;
• Integrated Octoparse AI scraper to identify and qualify leads at scale;
• Set up and managed Instantly.ai cold email campaigns with automated triggers;
• Connected positive email replies to personalized Retell AI-powered voice calls, enabling real-time follow-ups;
• Implemented automation to book appointments directly into Airtable CRM;
• Developed AI simulations and regression datasets for pre-deployment testing;
• Monitored and debugged live calls, applying prompt fixes and workflow optimizations.

Built an end-to-end MLOps integration with AWS SageMaker and Comet, a comprehensive model evaluation platform.

• Architected the full SageMaker/Comet MLOps integration, enabling seamless experiment tracking, model registry, deployment, and monitoring;
• Automated experiment logging and artifact versioning during SageMaker training, ensuring reproducibility and governance;
• Built one-click deployment workflows via AWS Lambda and dockerized runtimes, reducing dev to prod errors;
• Implemented production monitoring by streaming SageMaker inference logs into Comet MPM, providing teams with real-time visibility at scale;
• Streamlined the ML lifecycle for enterprise teams, reducing errors and enabling CI/CD-style automation.