Cat Y.

AI agent projects are easy to demo and hard to run. I am an AI engineer, and mine are in production at a Big 4 accounting firm, a 5,000-employee US enterprise, and a family office. 𝗘𝘅𝗽𝗲𝗿𝘁-𝗩𝗲𝘁𝘁𝗲𝗱 on Upwork, top 𝟭% of all freelancers platform-wide $𝟮𝟬𝟬𝗞+ earned as an AI engineer across AI agent, enterprise RAG, and LLM engagements, 𝗧𝗼𝗽 𝗥𝗮𝘁𝗲𝗱 𝗣𝗹𝘂𝘀, 𝟭𝟬𝟬% JSS 1,200+ hours logged on Upwork, most of it hands-on AI engineer work on AI agents, RAG, and LLM fine-tuning Founder of Super Cat Technology, an AI engineer team across 𝗛𝗼𝗻𝗴 𝗞𝗼𝗻𝗴 and 𝗟𝗼𝗻𝗱𝗼𝗻 building AI agent systems Feedback from an AI agent advisory and architecture engagement: "Very highly recommend working with Cat. The quality of work, going the extra mile and idea or assistance given were truly outstanding. We will continue to work with Cat on many projects." (ML Advisor and CTO Roadmap client, 2025 Upwork engagement) I build AI agent systems that survive real traffic, and every AI agent ships with evaluations, regression tests, and observability. I am Cat, an AI engineer, founder, and Fractional CTO at Super Cat Technology. As an AI engineer, my focus is production work: AI agent development, enterprise RAG over messy internal documents, LLM fine-tuning on domain data, and Claude and open-model deployments running 24/7. My job is to close the gap between an AI agent prototype and a system your operations team trusts at 2am. 💎 𝗦𝗘𝗟𝗘𝗖𝗧𝗘𝗗 𝗪𝗢𝗥𝗞 𝗕𝗶𝗴 𝟰 𝗮𝗰𝗰𝗼𝘂𝗻𝘁𝗶𝗻𝗴 𝗳𝗶𝗿𝗺. Enterprise RAG chatbot for junior auditors, multimodal RAG with OCR and computer vision over compliance libraries and audit guidelines. 𝗨𝗦 𝗰𝗹𝗶𝗲𝗻𝘁, 𝟱,𝟬𝟬𝟬+ 𝗲𝗺𝗽𝗹𝗼𝘆𝗲𝗲𝘀. A 24/7 AI agent call centre built on conversational AI, with telephony, appointment booking, and tool-calling through API integration into live systems. 𝗙𝗮𝗺𝗶𝗹𝘆 𝗼𝗳𝗳𝗶𝗰𝗲. Fine-tuned LLM with RAG over financial research and portfolio holdings, powering an AI agent investment workflow deployed on-prem and air-gapped. 𝗖𝗼𝗺𝗺𝗲𝗿𝗰𝗶𝗮𝗹 𝗽𝗿𝗼𝗱𝘂𝗰𝘁 𝗯𝘂𝗶𝗹𝗱, $𝟯𝟳𝗞 𝗼𝘃𝗲𝗿 𝟮𝟰𝟳 𝗵𝗼𝘂𝗿𝘀. Voice AI agent for a commercial voice product, 5-star client feedback. 𝗣𝗿𝗼𝗽𝗲𝗿𝘁𝘆 𝘁𝗲𝗰𝗵 𝗦𝗮𝗮𝗦. Natural language processing layer that turns plain-English property requests into exact search queries. 𝗢𝗽𝗲𝗻 𝘀𝗼𝘂𝗿𝗰𝗲. N8N2MCP, an MCP server generator for Claude with 𝟭𝟯𝟭 GitHub stars. ✔️ 𝗪𝗛𝗔𝗧 𝗜 𝗗𝗢 - AI agent development on LangGraph, LangChain, AutoGen, CrewAI, and SmolAgents, plus MCP servers wiring each AI agent into live business systems - Enterprise RAG engineering: multimodal RAG pipelines over compliance libraries and internal knowledge bases, with OCR and table parsing for semi-structured documents - Generative AI and AI integration: Claude, GPT, and open models behind one AI agent interface, with routing, caching, and cost control - LLM fine-tuning and deep learning on open models (Llama, Qwen 2.5, Phi, Deepseek, LLaVA) to ground AI agent behaviour in your domain language and data - Private and on-prem AI agent deployment for regulated data, including air-gapped environments - Conversational AI and voice agent development on LiveKit and similar stacks, with telephony, AI agent tool-calling, and 24/7 reliability - AI chatbot builds for customer service and internal knowledge - GraphRAG and LightRAG on Neo4J where vector search alone is not enough - Computer vision and enterprise OCR (CLIP, LLaVA, PaddleOCR, YOLO) feeding AI agent document understanding - Machine learning and data science for classification, extraction, and evaluation pipelines - Fractional CTO advisory: AI agent architecture review, evaluation strategy, and AI engineer team build 🎯 𝗪𝗛𝗢 𝗜 𝗪𝗢𝗥𝗞 𝗪𝗜𝗧𝗛 - Enterprise teams putting an AI agent into customer service, sales, or internal operations, not a one-off pilot - Financial services, family offices, Web3, supply chain, and regulated organisations with real data and real stakes - Funded startups that need an AI engineer to move an AI agent MVP into production - Teams that want a senior AI engineer on AI agent architecture and evaluation, not the cheapest pair of hands 🔄 𝗛𝗢𝗪 𝗜 𝗪𝗢𝗥𝗞 Step 1. Discovery: your use case, current stack, data reality, and what the AI agent has to get right in production. Step 2. Architecture: AI agent flow, tool boundaries, fallback behaviour, and an evaluation plan, delivered within a week. Step 3. Build and ship: a production AI agent with evaluations, regression tests, and LLMOps tooling behind it. Step 4. Handover or stay: train your team, or keep me on as your AI engineer and Fractional CTO while the AI agent matures in the wild. Send me a message or an invite with your AI agent, enterprise RAG, or LLM project. You get a hands-on AI engineer on the build, not a handoff. I reply within a few hours, 7 days a week, across US, UK, and EU time zones.