Arnav G.

𝟱𝟬𝟬,𝟬𝟬𝟬 𝗺𝗲𝘀𝘀𝗮𝗴𝗲𝘀/𝗺𝗼𝗻𝘁𝗵 𝗶𝗻 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝗼𝗻 𝗰𝗼𝗻𝘃𝗲𝗿𝘀𝗮𝘁𝗶𝗼𝗻𝗮𝗹 𝗔𝗜 | 𝟮𝟬,𝟬𝟬𝟬+ 𝗰𝗼𝗻𝗰𝘂𝗿𝗿𝗲𝗻𝘁 𝘂𝘀𝗲𝗿𝘀 𝗶𝗻 𝗿𝗲𝗮𝗹-𝘁𝗶𝗺𝗲 𝗶𝗻𝗳𝗿𝗮𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲. I build AI agents, automation, and the execution systems behind them, from product idea and architecture through implementation and production. ⭐⭐⭐⭐⭐ “He has a rare ability to balance startup agility with enterprise-grade reliability. He quickly identified potential monolithic bottlenecks, mapped out a highly scalable distributed system, and provided a crystal-clear technical blueprint for our internal engineering team to execute.” — Tyler, distributed AI backend architecture 𝗪𝗛𝗔𝗧 𝗜 𝗕𝗨𝗜𝗟𝗗 • Conversational AI: Owned engineering for a platform processing approximately 500,000 messages per month on Cloudflare. Worked directly with customers to understand their workflows and shape the product. • Core execution engines: Built and maintained a platform’s central agent execution engine, then rebuilt it around a skill-based architecture for orchestrating and carrying out tasks. • Multi-agent systems: Built orchestration, planning, computer-use, and observability components for the CustomGPT agent system. • Real-time infrastructure: Built WebRTC infrastructure supporting 20,000+ concurrent users using mediasoup, AWS, and Redis. 𝗦𝗢𝗠𝗘 𝗡𝗨𝗠𝗕𝗘𝗥𝗦 • 91.40 on LongMemEval, achieved in evaluations of an independently developed memory system combining session retrieval, classification, and planning/answering. • 93.36% on GAIA: The CustomGPT team’s reported score in its June 5, 2026 #1 leaderboard snapshot, with my contributions covering the agent harness and orchestration. 𝗪𝗛𝗔𝗧 𝗬𝗢𝗨 𝗖𝗔𝗡 𝗕𝗥𝗜𝗡𝗚 𝗠𝗘 A new AI product idea to shape and build from scratch. A platform that needs its execution engine. A workflow to automate. Or an existing agent that loses context, chooses the wrong tools, or fails to finish tasks. 𝗜 𝗖𝗔𝗡 𝗛𝗘𝗟𝗣 𝗪𝗜𝗧𝗛 • AI products: Product scoping, architecture, and implementation. • Agent orchestration: Tool use, task routing, state management, and multi-step workflows. • Retrieval and memory: Connecting relevant context to planning and answering. • Evaluation, debugging and tests: Reproducing failures and measuring improvements. • Production engineering: APIs, databases, integrations, and infrastructure. My core stack includes TypeScript, Node.js, Python, Cloudflare Workers, LangGraph, CrewAI, Jev and SQL databases. Framework choice follows your project’s requirements. ⭐⭐⭐⭐⭐ “Arnav’s ability to solve complex technical challenges and his dedication to delivering high-quality code were truly impressive.” — Fabio, WebRTC development project Send me your product idea, the workflow you want to automate, or a task your agent keeps failing. Let’s define what to build and what successful execution should look like.