{"header":{"name":"Cole McIntosh","title":"AI Engineer & Founder, Mellow AI","contact":{"github":"https://github.com/colesmcintosh","linkedin":"https://www.linkedin.com/in/colemcintosh/","location":"Dallas, TX","email":"cole@staymellow.ai"}},"summary":"AI engineer and founder of Mellow AI building production agent systems, document extraction pipelines, and developer tools in Python, TypeScript, and Rust. Combines hands-on systems work—from fine-tuned small models and zero-dependency parsers to multi-provider LLM infrastructure—with end-to-end delivery that measurably reduces operational load.","keyFacts":[{"label":"Role","value":"AI engineer; founder of Mellow AI"},{"label":"Based in","value":"Dallas, Texas, USA"},{"label":"Company","value":"Mellow AI (founded 2025), staymellow.ai"},{"label":"Focus","value":"Production AI agents, document extraction, model fine-tuning, and developer tooling"},{"label":"Languages","value":"Python, TypeScript, Rust, SQL"},{"label":"Notable work","value":"Core extraction platform for a global law firm (95% lower OCR cost, 75% lower latency); lease extraction API over 250K+ documents (<1% human correction)"},{"label":"Open source","value":"Merged PRs across LiteLLM (SSO, guardrails, OpenTelemetry, cost tracking, providers), Pydantic AI (xAI search/attachments, token counting), and Hugging Face smolagents"},{"label":"Contact","value":"cole@staymellow.ai"}],"skills":[{"category":"Languages","items":"Python, TypeScript, Rust, SQL"},{"category":"AI & Agents","items":"Agent harnesses, LLM evals & calibration, LoRA fine-tuning (Qwen), RAG, MCP, structured outputs"},{"category":"Frameworks & SDKs","items":"Pydantic AI, Vercel AI SDK, LangChain / LangGraph, LiteLLM, Zod, Next.js, Ratatui"},{"category":"Data & Cloud","items":"Custom OCR & PDF parsing, document forensics, OpenTelemetry & Langfuse, SQLite, AWS, Azure, GCP"}],"experience":[{"title":"Founder","company":"Mellow AI","period":"2025 – Present","description":["Re-architected the core document extraction platform and UI for a global law firm under strict data boundaries, modernizing ingestion and schema-validated pipelines.","Engineered a custom OCR and extraction pipeline cutting processing cost by 95% and latency by 75%.","Built and shipped Openextract (v1.0 on PyPI, plus a full TypeScript port)—a multimodal extraction toolkit supporting 10+ model providers with Pydantic/Zod validation.","Fine-tuned and released mev (LoRA on Qwen3.5-4B), an open-weight structured decision model reaching 86.7% dev accuracy with calibrated confidence (ECE 0.038)."]},{"title":"AI & Data Engineer","company":"Pay Ready","period":"2023 – 2025","description":["Architected a general-purpose support agent harness steered by composable prompts and skills, enabling rapid rollout of new operational workflows without bespoke pipelines.","Replaced a failing spam filter with a structured binary LLM classifier in live ticketing, cutting support agent workload by ~60%.","Shipped a production GPT-4 Vision lease and contract extraction API over 250K+ documents (<1% human correction).","Built an internal Slack RAG assistant adopted daily across teams in place of a bought tool, cutting manual effort by 60% and retrieval time by 75%."]},{"title":"Scaled Tooling Analyst","company":"Otter / CloudKitchens","note":{"label":"now Atoms","href":"https://atoms.co"},"period":"2022 – 2023","description":["Embedded with operations and engineering to build internal Python data pipelines scoring ~500GB/week.","Designed and shipped an internal Jira ticketing chatbot from discovery to production, cutting ticket resolution time by 30%."]},{"title":"Business Analyst","company":"RealPage","period":"2021 – 2022","description":["Built SQL data models and analytics dashboards supporting 75+ internal users, reducing manual reporting effort by 25%."]}],"openSource":[{"name":"LiteLLM","meta":"LLM Gateway · Python","description":"Shipped SSO admin API and UI, Google Cloud Model Armor guardrails, Langfuse OpenTelemetry tracing, usage/health dashboards, provider integrations (Hyperbolic, Lambda AI), Gemini caching and Perplexity search cost accounting, and Ollama streaming fixes.","pdfDescription":"SSO admin API/UI, GCP Model Armor guardrails, Langfuse OTel tracing, usage/health dashboards, new providers, caching/search cost tracking, and Ollama stream fixes.","link":"https://github.com/BerriAI/litellm/pulls?q=is%3Apr+is%3Amerged+author%3Acolesmcintosh"},{"name":"Pydantic AI","meta":"Agent Framework · Python","description":"Implemented xAI search tools and attachment lifecycles, model-agnostic X search via subagent fallback, OpenAI input token counting, UI adapter URL sanitization, and provider host/timeout settings.","pdfDescription":"xAI search tools & attachment lifecycles, subagent X search fallback, OpenAI input token counting, UI URL sanitization, and provider host/timeout config.","link":"https://github.com/pydantic/pydantic-ai/pulls?q=is%3Apr+is%3Amerged+author%3Acolesmcintosh"},{"name":"smolagents","meta":"Hugging Face · Python","description":"Fixed Enum and custom metaclass execution in sandboxed agent code, added AWS Bedrock API-key support, and refactored MultiStepAgent test coverage.","pdfDescription":"Fixed Enum/metaclass sandbox execution, added AWS Bedrock API-key support, and refactored MultiStepAgent test suite.","link":"https://github.com/huggingface/smolagents/pulls?q=is%3Apr+is%3Amerged+author%3Acolesmcintosh"}],"projects":[{"name":"TeraCode","meta":"Rust · Multi-agent TUI","description":"Local-first Ratatui switchyard and bounded DAG scheduler over 6 coding-agent CLIs with git worktree isolation, binary-safe patch assembly, SQLite run history, and explicit autonomy gates.","pdfDescription":"Local-first Rust TUI & bounded DAG scheduler over 6 coding-agent CLIs with git worktree isolation, binary-safe patching, SQLite history, and autonomy policies.","link":"https://github.com/colesmcintosh/teracode"},{"name":"pdfmd","meta":"Rust · Zero-dep Engine","description":"Zero-dependency PDF-to-Markdown engine built from scratch—PDF parser, DEFLATE decoder, font decoding, and layout reconstruction exposed via CLI and Python bindings.","pdfDescription":"Zero-dependency Rust PDF-to-Markdown engine from scratch (PDF parser, DEFLATE decoder, font & layout reconstruction) with CLI and Python bindings.","link":"https://github.com/colesmcintosh/pdfmd"},{"name":"Openextract","meta":"Python · TypeScript","description":"Multimodal extraction toolkit converting documents, images, audio, and video into schema-validated Pydantic/Zod objects across 10+ model providers.","pdfDescription":"Model-agnostic Python/TypeScript extraction toolkit turning documents and media into validated Pydantic/Zod objects across 10+ providers.","link":"https://github.com/Mellow-Artificial-Intelligence/openextract"},{"name":"Liedar","meta":"TypeScript · Forensics","description":"Document tampering screening library, CLI, and browser UI combining file metadata, copied-region image detection, and model consistency assessments for human review.","pdfDescription":"Document tampering screening library, CLI, and browser UI combining metadata, copied-region detection, and model assessments for human review.","link":"https://github.com/Mellow-Artificial-Intelligence/liedar"},{"name":"mev","meta":"LoRA · Qwen3.5-4B","description":"Open-weight 4B decision model fine-tuned for 2–24 option structured state selection with calibrated confidence; weights, training notes, and evaluation published on Hugging Face.","pdfDescription":"Open-weight Qwen3.5-4B LoRA decision model for 2–24 option structured state selection with calibrated confidence (Hugging Face).","link":"https://huggingface.co/stay-mellow-ai/mev"}],"education":["RAG, LangChain, LangGraph & Semantic Search — DeepLearning.AI","Azure Applied AI — Microsoft","AWS Cloud Technical Essentials — Amazon","Google Cloud Fundamentals — Google"]}