Tibor Thompson

Full-Stack AI Engineer | Founding Team Experience | Stanford CS '24

Full-stack software engineer specializing in AI infrastructure and production LLM applications. Early engineer at venture-backed AI startups, shipping agentic platforms and RAG systems from concept to scale.

Get in Touch
Call Orbit

Chat with Orbit, my AI agent, in the bottom right or call +1 (877) 344-0719

Agentic AI Multi-Agent Systems RAG Systems Production LLMs

Work Experience

ProphecyGov

Software Engineer

2025
  • ▹ Architected and shipped AI research citation engine with self-evaluating confidence scores and fuzzy matching algorithms
  • ▹ Built LLM-as-judge evaluation system (inspired by Anthropic methodology) for automated quality assessment of orchestrator agent
  • ▹ Engineered SQL-based agent action framework enabling complex database operations through natural language interface
  • ▹ Optimized vector search with advanced filtering systems, improving retrieval precision across large document repositories
  • ▹ Led technical demonstrations at industry conferences

Vanilla AI

Software Engineer (Founding Team)

2024-2025
  • ▹ Sole engineer on founding team: architected and shipped complete AI receptionist platform from concept to production
  • ▹ Built scalable AWS infrastructure with CI/CD pipelines and comprehensive testing, maintaining 24/7 uptime for live customer-facing AI agents
  • ▹ Designed and launched contractor-matching MVP; presented technical architecture to investors as founding technical lead
  • ▹ Launched production platform with live AI agents handling voice calls and payment processing

Stanford University

AI Research Contributor

2023-2024
  • ▹ Published research at NeurIPS on automating digital workflows using Vision-Language Models (VLMs)
  • ▹ Contributed to dataset generation, testing, and model validation
  • ▹ Project focused on developing AI tools for enterprise workflows using the WONDERBREAD benchmark

Kaladin Inc.

Software Engineer Intern

2023
  • ▹ Designed and implemented cryptocurrency algorithmic trading system with supervised learning
  • ▹ Conducted backtesting simulations on historical data to refine strategies for testnet deployment
  • ▹ Integrated REST APIs for real-time market data and built front-end visualization tools
  • ▹ Represented company at 3 industry conferences, gathering user feedback to inform product roadmap

Selected Work

Project snapshots focused on the problem, technical ownership, and measurable outcomes.

Municipal AI Research Platform

ProphecyGov, founding engineer

2025

Problem

Municipal teams need trustworthy AI answers with source attribution across large government document repositories.

Role

Architected citation, retrieval, and evaluation systems for an agentic local-government platform.

Outcome

Supported technical demos at ICMA 2025 that contributed to 200+ qualified municipal contract leads.

Technical Approach

  • Built a research citation engine with confidence scoring and fuzzy source matching
  • Implemented LLM-as-judge evaluation flows for orchestrator quality checks
  • Engineered SQL-backed agent actions for natural language database operations
Python LLM-as-judge Vector Search SQL Agents RAG

Production AI Receptionist

Vanilla AI, founding team

2024-2025

Problem

Real estate and home-service businesses needed automated customer intake, voice handling, and proposal workflows.

Role

Sole engineer responsible for full-stack architecture from concept through production launch.

Outcome

Launched live customer-facing AI agents handling voice calls and payment flows.

Technical Approach

  • Built voice-agent workflows, payment processing, and automated proposal generation
  • Shipped AWS infrastructure, CI/CD, and production deployment workflows
  • Presented technical architecture to investors as founding technical lead
Voice AI AWS CI/CD Payments Full-stack

WONDERBREAD Workflow Automation Research

Stanford Hazy Lab

2023-2024

Problem

Digital business workflows are difficult for multimodal agents to evaluate and automate reliably.

Role

Contributed to dataset generation, testing, and model validation for VLM-based workflow automation.

Outcome

Helped publish research on automating digital workflows with Vision-Language Models.

Technical Approach

  • Validated multimodal foundation models on business-process tasks
  • Contributed testing workflows and benchmark data quality checks
  • Supported research published at NeurIPS 2024
VLMs Benchmarking Model Validation Python Research

Generative Email Client

Stanford senior project

2023-2024

Problem

Users need email drafts that reflect personal context and communication preferences.

Role

Built a RAG email client using OpenAI and Meta Llama models with Stanford AI researcher collaboration.

Outcome

Created a beta-published Chrome extension for context-aware email drafting.

Technical Approach

  • Implemented retrieval over user context for personalized draft generation
  • Compared OpenAI and Llama model behavior for generation quality
  • Prepared the extension for Chrome Web Store beta publication
OpenAI Meta Llama RAG Chrome Extension TypeScript

Court-Of-Agents

Personal project

2024

Problem

Multi-agent debates are hard to inspect when agents only produce isolated text responses.

Role

Built a voice-enabled multi-agent simulation for courtroom-style debate and resolution.

Outcome

Demonstrates practical multi-agent orchestration beyond static chat interfaces.

Technical Approach

  • Designed agent roles for collaborative argumentation and adjudication
  • Integrated voice interaction patterns for live simulation flow
  • Added repeatable test coverage around agent configuration
Multi-agent Systems Voice AI Python Testing

Skills & Technologies

Backend & Infrastructure

Python Flask FastAPI PostgreSQL REST APIs AWS EC2 Docker GitHub Actions

Frontend

Next.js TypeScript React.js JavaScript

AI & Machine Learning

OpenAI LangGraph Google Gemini Pinecone PyTorch TensorFlow RAG Multi-agent orchestration LLM-as-judge evaluation NLP Supervised Learning Reinforcement Learning Graph Neural Networks

Cloud & DevOps

AWS Google Cloud Automated testing Observability Production deployment CI/CD

Additional Languages

C++ C Solidity Flutter/Dart SwiftUI

Get in Touch

Email

tibor@tiborthompson.com Primary
tethomp@alumni.stanford.edu

Location

San Francisco, California

LinkedIn GitHub