Steve Leve
Full-Stack AI Engineer
I build products and the agentic systems that build them.
Production software and the agentic development systems behind it: coding-agent harnesses, context engineering, knowledge architecture, agent memory, and verification workflows for AI-generated code. Backed by 25 years of shipping and modernizing systems at scale.
What I Do
Full-Stack AI Engineering
Customer-facing AI agent services, production RAG, and multi-LLM features. I build AI applications that ship—with sub-second latency, streaming responses, and the evaluation frameworks to prove they work.
Agentic Development Systems
I build the systems that build the software: coding-agent harnesses, cross-harness synchronization, agent memory, and semi-autonomous agent teams—with the verification discipline that makes agent-driven development reliable.
AI-Augmented Legacy Modernization
25 years inside complex production systems, 17 of them at $1B+ transaction scale. I use custom sub-agents to analyze and migrate legacy subsystems—modernization without breaking what works.
Live Demos
See the work, not just the words.
Vercel RAG Demo
Production streaming chatbot with semantic search. Next.js 16, React 19, OpenAI GPT-4o, Neon PostgreSQL (pgvector). Sub-second responses across 100+ document corpus.
Visit Demo →Cloudflare Workers RAG Demo
Edge-deployed RAG with on-device LLM inference (Llama-3.1-8B). Zero external API calls, optimized for global latency. Built on Cloudflare Workers, Vectorize, D1, R2.
Visit Demo →Recent Work
PartnershipTool (Cofounder, Ongoing)
Ad-tech business intelligence platform (Next.js, React, PostgreSQL) built with an agent-driven workflow: Cursor + Claude, self-hosted Honcho for agent memory, and Hermes as a semi-autonomous agent team.
Customer AI Agent Service
Co-authored a customer-facing account management AI agent (AWS Bedrock, LangChain, Temporal), plus multi-agent knowledge architecture and cross-harness agent synchronization for Cursor, Copilot, and Claude Code.
RAG Evaluation Methodology
Metrics-driven RAG quality improvement using Precision@K, Recall, MRR, NDCG. Achieved +15% precision and +22% overall quality through systematic experimentation.
Let's Talk
I'm looking for senior IC roles on small, fast-moving, high-ownership teams working at the frontier of agent-driven development. Remote only.