Architecting and shipping production-grade Generative AI systems and
agentic workflows - embedded with enterprise teams, from prototype to scale.

I'm Jin, a Senior Lead Forward Deployed Engineer on Google's Generative AI team. Based in New York, I specialize in the architecture, deployment, and operation of production-grade Generative AI systems - working at the frontier of agentic workflows, multi-agent orchestration, foundation model fine-tuning, and enterprise-scale ML systems.

As an FDE, I embed directly with customer engineering teams to design, code, and ship bespoke GenAI applications that drive measurable business impact - building the connective tissue between Google's AI products and real-world infrastructure, data, and security perimeters. My focus is taking ambitious ideas from rapid prototype to hardened, observable, production-grade reality.

Before Google, I was a Member of Technical Staff - Forward Deployed Engineer at TwelveLabs, a frontier video-understanding lab building state-of-the-art Vision-Language Models (VLMs). There I worked hands-on with customers to deploy multimodal video AI - semantic search, understanding, and retrieval - into production at scale.

Earlier, I was a Senior GenAI Engineer at Amazon Web Services within the Global Prototyping team, where I led the design of AI-driven applications for complex enterprise challenges. And as a Research Scientist at the Air Force Research Lab, I worked at the intersection of applied research and mission-critical systems - bridging cutting-edge R&D with scalable, real-world execution.

• 🔭 Senior GenAI Forward Deployed Engineer @ Google - building and shipping production Generative AI applications, agentic systems, and multi-agent workflows
• 🎥 Previously Member of Technical Staff, Forward Deployed Engineer @ TwelveLabs - deploying frontier Vision-Language Models (VLMs) for video understanding
• 🏢 Formerly Senior GenAI Engineer @ Amazon Web Services (Global Prototyping)
• 🇺🇸 Former Research Scientist @ Air Force Research Lab
• 📄 Published & presented at ICML 2025
• 🌱 10x AWS Machine Learning Certified
• 📇 Speak three languages - Spanish, Chinese, and English
• 💻 Personal Website
• 📝 Resume

Mood Swings: Three Neuromodulatory Scalars Drive Impulse–Caution Shifts in Deep Actor–Critic Agents - ICML 2025

This work introduces a novel framework for dynamic emotional states in multi-agent systems, where deep reinforcement learning agents experience and adapt to "mood swings" driven by environmental stimuli and inter-agent interactions. Modeled through three neuromodulatory scalars, these states shift agents along an impulse–caution spectrum - demonstrating how emotional volatility can sharpen decision-making and improve collaborative outcomes in complex multi-agent scenarios.