Selected AI Engineering Projects
A selection of AI products and systems I’ve built for clients, from LLM applications to production backend infrastructure.
- Python
- FastAPI
- Asyncio
- AWS EC2
- CI/CD
- AI Agents
- RAG
- OpenAI API
- REST API
- WebSockets
AI BACKEND · AGENT ARCHITECTURE
Production AI backend for SalesGo
A production AI backend powering conversational agents, coaching, RAG, prompt orchestration, and real-time voice experiences.
Problem
SalesGo needed a reliable backend to support multiple interconnected AI systems, concurrent LLM workflows, coaching features, and real-time voice interactions.
My work
Architected and built the AI backend in Python and FastAPI, including Asyncio orchestration, REST APIs, AWS EC2 deployment, CI/CD, RAG systems, and real-time voice infrastructure.
Outcome
Delivered the backend and orchestration layer powering the platform’s AI processing, including suggestions, scoring and feedback chains, AI coaching, and conversational voice role-play.
- Python
- GPT-4
- Prompt Engineering
- AI Agent Development
- OpenAI API
- AI App Development
- ChatGPT Integration
- LLM Implementation
- ChatGPT
- Streamlit
AI AGENTS · CUSTOM SALES COACHING
Custom GPT-4 sales coach
A GPT-4-powered coaching system tailored to a client’s proprietary sales process and methodology.
Problem
Generic AI-generated coaching did not consistently reflect the client’s unique sales methods or different sales scenarios.
My work
Designed the AI agent and prompt framework, including response classification and dynamic guidance adjustments, using Python, GPT-4, and Streamlit.
Outcome
Delivered more relevant, context-aware coaching aligned with the client’s sales process across a variety of situations.
- Prompt Engineering
- AI Agents
- Python
- ChatGPT
- OpenAI API
- FastAPI
- LLM Implementation
- AI App Development
- Python
- REST API
AI AGENTS · SALES ENABLEMENT
Conversational AI sales role-play evaluator
An AI-powered sales coaching application that evaluates role-play conversations, objection handling, and appointment-setting performance.
Problem
Sales teams needed a consistent way to evaluate practice conversations and the effectiveness of objection handling.
My work
Built the AI-agent and prompt-engineering workflow with a Python, FastAPI, and REST API backend for analyzing role-play conversations.
Outcome
Automated feedback on objection handling, conversation performance, and whether the agent successfully secured an appointment.