Shumeng Liu

I transitioned from finance into AI product development, focusing on AI chat applications, Agent workflows, and multimodal product delivery. I’ve built two AI products from 0→1 using Prompt, RAG, and Multi-Agent architectures, with hands-on experience across mainstream LLMs.

I move quickly from idea to MVP—using Vibe Coding and AI coding tools like Cursor/Codex to validate value early. I emphasize end-to-end ownership: requirement discovery, system prompt tuning, integrating RAG memory, and iterating based on user feedback.

Before AI, I spent four years in finance and business analysis, including system implementation support (HFM→EPM), forecasting, and data-driven decision support. That foundation helps me treat AI features as measurable experiences—balancing efficiency with quality, and using evaluation outputs (e.g., persona consistency) to guide iteration.