​Summary

​This new session of Applied ML & AI (AMA) is on "Deploying LLMs in the Real World", highlighting real-world agent deployments, from developer workflows and enterprise knowledge systems to IP innovation and human-computer interaction patterns.

​​Schedule

​18:00 Doors open

​18:30 - 20:00 Speaker sessions

​20:00 - 21:00 Networking

​21:00 Event ends

​Talks

​Talk 1: Maximizing AI coding assistants as a software engineer

​Speaker: J. Mario Meissner, ML Engineer, Indeed

​AI Coding tools are growing in popularity. We've all tried GitHub Copilot, and some of us have played with Cursor. But as a professional software engineer, it's easy for these tools to fall short of our expectations, often messing things up more than actually helping out. This talk covers how to actually maximize the value of these tools, establishing guardrails, improving context management, and showcasing advanced workflows. By the end of this talk, you too can be a 10x engineer with the power of AI.

​Talk 2: Transforming IP & Innovation Workflows by leveraging LLMs & Generative AI

​Speaker: Evenpreet Singh, PM, XLSCOUT

​Evenpreet Singh from XLSCOUT explores how generative AI, especially custom LLMs, is revolutionizing intellectual property (IP) workflows. From ideation to patent drafting, the talk highlights how proprietary tools like ParaEmbed, ParaRAG, and ParaRANK are driving smarter, faster IP decisions. Beyond showcasing these technologies, Evenpreet will share insights from academic collaborations and explain how domain-specific AI improves accuracy and strategic outcomes across R&D, legal, and innovation teams. This session is a look at the future of IP when deep expertise meets powerful, tailored AI.

​Talk 3: Implementing Complex RAG Ops Systems with the LLMStack Framework

​Speaker: Sigrid Jin, ML DevRel, Sionic AI

​This session demonstrates how to construct an enterprise-grade RAG Ops foundation using Sionic AI’s STORM Platform. Beginning with data ingestion and embedding generation, we will show how to connect a vector database, assemble retrieval-and-generation chains, and deploy inference endpoints, all within a unified console. Live demos cover real-time evaluation metrics, guardrail policies, and the latest no-code UI that speeds configuration and governance. Attendees will leave with a practical roadmap for moving from PoC to full production, minimizing hallucinations while maximizing time-to-value for internal data. The talk targets engineers, MLOps specialists, and DX leaders seeking to operationalize RAG swiftly and securely in their organizations.

​Talk 4: Agentic Patterns in LLM

​Speaker: Venali Sonone, Manager, American Express

​This talk explores Agentic Patterns in Large Language Models (LLMs), focusing on emerging design strategies that enable LLMs to exhibit autonomous, goal-directed behavior. We will examine key architectural principles, prompting techniques, memory systems, and coordination mechanisms that empower LLMs to function as agents. Real-world applications and research directions will be highlighted to demonstrate how agentic patterns can unlock more dynamic, interactive, and context-aware AI systems.

​Speaker Bios

​J. Mario Meissner
Mario is an LLM researcher and AI engineer specializing in AI coding assistants. He holds a degree from the University of Tokyo and currently works as an ML Engineer at Indeed Japan. Based in Tokyo, he focuses on advancing developer productivity through AI-powered tooling.

​Evenpreet Singh
Evenpreet is a Product Manager at XLSCOUT with over six years of experience in the IP industry. His work spans IP research, product management, R&D, and customer success. He collaborates globally with IP and R&D leaders to develop high-impact, AI-driven innovation strategies.

​Sigrid Jin
Sigrid is a Machine Learning DevRel Engineer at Sionic AI, leading developer engagement across Korea, Japan, and Southeast Asia. With over three years of experience as an ML and Python backend engineer, he bridges technical expertise with regional business strategy.

​Venali Sonone
Venali is an AI and data science leader with more than a decade of experience building scalable, intelligent systems. She works at the intersection of business and technology, with deep expertise in LLMs, generative AI, and responsible AI development. A frequent speaker, she focuses on creating real-world impact through thoughtful AI design.

​Sponsor

​Code Chrysalis is a premier coding bootcamp based in Tokyo, dedicated to transforming individuals into world-class software engineers. Founded in 2017, the program offers a rigorous, full-time immersive curriculum, focusing on full-stack modern software development practices and career readiness.

​​Our Community

​Tokyo AI (​​TAI) is the biggest AI community in Japan, with 2,400+ members mainly based in Tokyo (engineers, researchers, investors, product managers, and corporate innovation managers).