Quancheng Vincent Li

About Quancheng Vincent Li

I’m an AI/Software Engineer with a background in full-stack development, applied machine learning, and intelligent systems design. Currently pursuing my M.S. in Computer Science at Northeastern University (Seattle)(GPA: 3.97), I specialize in building AI-driven platforms that bridge technical innovation with real-world business impact. AI & Product Engineering – At VSDsign, I engineered an AI Agent orchestration platform with n8n, Claude, and AWS that automated workflows from reporting to multimodal content generation. Applied Machine Learning – I’ve developed advanced NLP pipelines and RAG systems: from aviation document retrieval with GraphRAG + LangChain to e-commerce review authenticity detection, where my model achieved 98.8% accuracy and 0.979 ROC-AUC using BERT and GPT-2. Full-Stack & Cloud – I design scalable systems end-to-end: a subscription management platform with React, Node.js, PostgreSQL, and LangChain for spend analytics and AI-based advisors; and a distributed Java-based lift ride simulation API that handled 2.5K req/s on AWS with Terraform-provisioned infrastructure. I thrive at the intersection of AI, distributed systems, and product design, building tools that are not only technically robust but also enhance user experience and business value.I’m excited to apply my skills to AI/ML engineering, full-stack system design, and intelligent automation, especially in roles where I can ship impactful products, scale intelligent infrastructure, and collaborate with cross-disciplinary teams.

Experience

1. Software Engineer

   Casehero

   Nov 2025 — Present

   Optimized the LLM generation pipeline by migrating from monolithic text generation to a hybrid Nunjucks architecture. Restricted GPT strictly to structured entity extraction, eliminating hallucinations in legal boilerplate and slashing document generation latency from 4 minutes to under 1 minute.Engineered a deterministic document assembly engine that maps dynamic LLM outputs into Lexical rich-text ASTs and compiles these nodes directly into DOCX format, ensuring 100% formatting consistency between the web editor and final legal exports.Overhauled the frontend architecture using Shadcn UI, TanStack Query, and Compound Components. Integrated Zod validation and Clerk/Stripe to decouple UI from business logic, reducing technical debt by ~40% and enforcing strict data hygiene for downstream AI processing.

Education

• Suffolk University - Sawyer Business School

  Master of Science in Finance, Finance

• Northeastern University

  Master of Science - MS, Computer Science

• Suffolk University - Sawyer Business School

  undergraduated, Finance, General

  2018 — 2022