Thel Gunter

Resume

Applied AI engineer. Production LLM agent platforms end to end, LLMs integrated into real products, and a background in large-model inference and computer vision on edge hardware.

PDF on request at thelgunter21@gmail.com.

Experience

  1. Networking infrastructure company

    Jul 2026 to Sep 2026

    Contract AI Engineer / Remote

    • Built a production AI agent platform (Node.js / TypeScript, Anthropic Claude, Model Context Protocol) that lets customers manage real device infrastructure through natural-language chat, with streaming responses over GraphQL / WebSocket.
    • Engineered delegated per-user authorization for AI-driven infrastructure actions: OAuth2 PKCE with Ory Hydra, dynamic client registration, RFC 8707 audience-bound JWTs, and server-side JWKS verification, so the agent acts strictly as the authenticated user, with identity-stamped audit trails.
    • Designed human-in-the-loop safety controls: automatic read/write classification of 43 MCP tools, with mandatory user confirmation gating destructive operations.
    • Drove quality through test-driven development (80+ automated tests, including end-to-end OAuth flow tests against a mock identity provider). Root-caused a production auth incident via JWT forensics and hardened the tooling against that failure class.
    • Designed the service's AWS deployment architecture on ECS behind an Application Load Balancer.

  2. Civic engagement platform (confidential client)

    Oct 2025 to Mar 2026

    Contract Software Engineer / Remote

    • Architected and deployed a FastAPI / React platform on AWS (Elastic Beanstalk, S3, CloudWatch monitoring) with PostgreSQL and Redis for scalable letter generation and campaign management.
    • Integrated the OpenAI GPT-4o API for AI-powered letter personalization using a custom voice-profile system. Built an admin dashboard with analytics, user management, and campaign-coordination tools.

  3. Cachengo

    Jun 2024 to Jan 2026

    Lead, Software Integration / Huntingdon, TN

    Joined as Software Engineer. Promoted to Lead within two months.

    • Designed a decentralized computer-vision pipeline over an internal pub/sub system, architected to support 100K+ cameras streaming to the cloud, and built a low-resolution threat-detection model for real-time inference on constrained edge hardware.
    • Upgraded the production CV model from YOLOv5s to YOLOv8n, quantized it for NPU deployment, and rewrote post-processing to run efficiently on CPU, improving edge throughput.
    • Led a team building image-labeling workflows and dataset management for all deployed models.

  4. Capgemini

    Jun 2023 to Jun 2024

    Consultant / Columbia, SC

    • Built and presented five Azure OpenAI retrieval-augmented generation (RAG) proofs of concept to enterprise clients for knowledge management over proprietary documents, including document chunking, embedding generation, and vector search.
    • Delivered two customer-service AI POCs integrating Soul Machines' digital-person technology for voice-activated support.

Projects

Skills

AI / ML
LangGraph, LangChain, multi-agent systems, RAG (chunking, embeddings, vector search), Model Context Protocol (MCP), OpenAI and Anthropic APIs, quantization
Backend
FastAPI, Node.js, TypeScript, React, Next.js, PostgreSQL, Redis, Docker, REST APIs, GraphQL / WebSocket streaming, AWS (ECS, Elastic Beanstalk, S3, CloudWatch), Azure AI / Azure OpenAI
Languages
Python, JavaScript, TypeScript, C/C++ (embedded), SQL
Inference
AMD Instinct MI300X (ROCm / RCCL), Qualcomm AI 100 (NPU), tensor-parallel serving, edge inference
Vision
YOLO (v5 to v8), SAM / SAM 3D Body, MediaPipe, OpenCV, pose and keypoint estimation

Education and certifications

Degrees
University of South Carolina. B.S. Computer Science (Aug 2022). B.S. Business Administration, Finance (May 2020).
Certs
Microsoft Certified: Azure AI Fundamentals. Microsoft Certified: Azure Fundamentals.