Emil Ayvazyan

AI Engineer, Software Engineer at Freelance

1. Advanced Website Chatbots Built a chatbot for a website with 30,000+ products. It suggests products (shown as cards) and answers FAQs. Tech: Pinecone (vector embeddings), OpenAI Assistants API (multi-agent system logic with multiple GPTs for query rewriting, customer satisfaction tracking, and engagement). Impact: Improved product discovery and customer experience. 2. AI Voice Agent (Conversational Calling Agent) Developed a voice-calling agent for a US home repair service. It handles incoming calls, checks available times in the database, books appointments, collects customer info, and adds confirmed bookings to Google Calendar. Tech: bland.ai, Python (FastAPI), SQL (availability checks), Make.com (workflow automation). Impact: Automated booking process and reduced manual workload. 3. SaaS Application (eBay Fitment Generator) Created a SaaS app that helps sellers manage eBay listings by auto-generating product fitments. Users authorize their eBay account, the app generates fitments with OpenAI Assistants API, and then pushes them via the eBay API. Tech: OpenAI Assistants API, eBay API, Stripe (subscription payments). Impact: Saved sellers time and improved listing accuracy.1. Advanced Website Chatbots Built a chatbot for a website with 30,000+ products. It suggests products (shown as cards) and answers FAQs. Tech: Pinecone (vector embeddings), OpenAI Assistants API (multi-agent system logic with multiple GPTs for query rewriting, customer satisfaction tracking, and engagement). Impact: Improved product discovery and customer experience. 2. AI Voice Agent (Conversational Calling Agent) Developed a voice-calling agent for a US home repair service. It handles incoming calls, checks available times in the database, books appointments, collects customer info, and adds confirmed bookings to Google Calendar. Tech: bland.ai, Python (FastAPI), SQL (availability checks), Make.com (workflow automation). Impact: Automated booking process and reduced manual workload. 3. SaaS Application (eBay Fitment Generator) Created a SaaS app that helps sellers manage eBay listings by auto-generating product fitments. Users authorize their eBay account, the app generates fitments with OpenAI Assistants API, and then pushes them via the eBay API. Tech: OpenAI Assistants API, eBay API, Stripe (subscription payments). Impact: Saved sellers time and improved listing accuracy.

AI Engineer, Data Scientist at CareAware

1. Led the development of an Optimal Price Detection neural network for a billboard with 10000 assets nationwide. It predicts booking ratios based on location lighting size weekly trends and pricing to recommend optimal weekly rates and maximize revenue. As the project winner I achieved 97 percent prediction accuracy outperforming peers. Technical approach Built and trained a Neural Network model using PyTorch to capture temporal dynamics in historical data. Preprocessed features for example scaling prices encoding sizes normalizing lat long for robust predictions. Deployed model for evaluation on validation sets and final price recommendations. For details see GitHub [https://github.com/emil-ayv/Bootcamp-Final-Project-CAREAWARE]. 2. Designed multi agent systems for collaborative AI problem solving simulating real world scenarios where agents interact to achieve complex goals. Technical approach Utilized frameworks like AgencySwarm and CrewAI to orchestrate agents with specialized roles incorporating communication protocols and shared environments.

AI Engineer, Software Engineer at B2B Connection

1. Built an interactive website chatbots for customer support using Python (FastAPI) for backend logic, JavaScript for dynamic interactions, and Pinecone for data embeddings. Deployed on Replit. The bot handles FAQs, recommends packages (shown as cards with name, price, image, and description), and explains project milestones to improve customer engagement. 2. Developed a Store CRM system integrated with WordPress, featuring four user roles (Seller, Cashier, Manager, Admin). Backend in Python (FastAPI), frontend in JavaScript, and MySQL for database management. Deployed on Replit to streamline inventory, sales, and management workflows.

Built an interactive chatbot for a website with 30,000+ products. Suggested products (cards) and answered FAQs. Tech: Pinecone (vector embeddings), OpenAI APIs, multi-agent system logic for query rewriting and customer satisfaction.

Developed an AI voice-calling agent for a US home repairs service. Handles incoming calls, checks available appointment times, books confirmed appointments on Google Calendar. Tech: Python (FastAPI), PyTorch, SQL (availability checks), Make.com for workflow automation.

Created a SaaS app that helps sellers manage eBay listings by auto-generating fitments. Users authorize their eBay account; app generates fits via OpenAI API and the eBay API; payments via Stripe. Tech: OpenAI API, eBay API, Stripe. Impact: saved sellers time and improved listing accuracy.

Led development of an Optimal Price Detection neural network for a billboard with 10,000 assets nationwide. Predicts booking ratios based on location, lighting, size; recommends optimal weekly rates to maximize revenue. Tech: PyTorch; preprocessed features for pricing; deployed model for evaluation on validation sets and final price recommendations.