# Zhuolun Li — Software Engineer > Software engineer specializing in AI infrastructure, LLM systems, and developer tools. Currently building autonomous AI coding agents at Macroscope ($40M-funded), a San Francisco-based company focused on AI-powered code review. ## About Zhuolun Li is a software engineer specializing in AI engineering and AI coding agents. Currently at Macroscope ($40M-funded), he builds autonomous AI coding agents that leverage AST-based code analysis, knowledge graphs, and multi-model LLM pipelines to catch real bugs before they hit production. Previously, he built a GenAI platform at Fund Launch that automated fund creation workflows, helping facilitate over $4B+ in fund creation. Before that, he was at Alibaba Group (Lazada), engineering high-throughput reward systems serving 100M MAU across 12 countries at 180K QPS. At UC Berkeley RIOS Lab, he worked on LLVM instruction scheduling optimization for RISC-V architecture under Prof. David Patterson. He has won multiple hackathons building AI agents — from end-to-end testing bots to web security scanners and multi-player RPG game masters. He holds a Master's in Computer Engineering from NYU Tandon School of Engineering. --- ## Contact & Links - Email: lizhuolun.nyu@gmail.com - Website: https://www.zhuolunli.work/ - GitHub: https://github.com/icecreamlun - LinkedIn: https://www.linkedin.com/in/zhuolun-li-267757290/ --- ## Professional Experience ### Macroscope — Software Engineer (Dec 2025–Present, San Francisco) - Building best-in-class autonomous AI coding agents that plan, write, review, and iterate on real production code - Working with AST-based code analysis, multi-model frontier LLM pipelines, and knowledge graph systems ### ByteDance (Trae) — Fellow (Sep 2025–Present) - Hosted community events including workshops and hackathons for the Trae AI developer community - Shared expertise through demos, case studies, and tutorials ### Fund Launch — Gen AI Engineer (Mar 2025–Oct 2025, Lehi, Utah) - Led full-cycle development of a GenAI-powered fund creation platform, facilitating creation of funds totaling over $4B in managed scale - Designed a modular DAG-based task orchestration framework with agent-based control flow, supporting dynamic routing between LLM agents and function buckets - Built multiple RAG pipelines using Supabase + Qdrant + PostgreSQL full-text indexing, reducing average token usage by 50% - Implemented bucketed agent architecture serving 10K+ MAU with <2s latency - Fine-tuned custom LLMs on AWS SageMaker, reducing hallucination rate by 15%; deployed on AWS EC2 with Datadog and LangSmith monitoring ### Alibaba Group (Lazada) — Software Development Engineer (Jun 2024–Oct 2024, Guangzhou, China) - Worked on the Lazada e-commerce reward service platform supporting 12 countries, 100M MAU, and 180K QPS - Designed and implemented APIs to deploy bulk user tasks using internal RPC framework; launched one-click bulk update feature - Optimized local and Redis distributed caches achieving 98% cache hit rate and 20% latency reduction; reduced storage costs by 12% - Conducted system stress testing reaching 50K API peak QPS - Performed system version upgrades, optimized Jedis, enhanced translation engine with MQ, reduced log memory by 40% ### Freeleaps — Applied AI/ML Engineer Intern (Mar 2025–Jun 2025, Seattle, Remote) - Built a retrieval-based job recommendation engine using Flink, improving F1-score of top-10 recommendations by 18% - Developed MongoDB Kafka Connectors for near-real-time synchronization of transactional data across streaming pipelines - Architected a sub-200ms latency ETL pipeline using Kafka, Pinot, and Hive with Apache Flink for event-time window processing - Developed a Gitea-integrated AI code review bot using GPT-4o, FastAPI, and Kubernetes — reduced manual review time by 60% and post-merge bugs by 35% across 100+ contributors ### ResuMaster — Founding Full Stack Developer (May 2024–Jan 2025) - Co-built an AI-powered resume editing platform — 70K+ resumes processed with 98% user satisfaction - Built RESTful APIs with Python/Flask, integrated OpenAI GPT-4o for chat-based resume editing - Developed a Chrome extension for auto job fetching and form filling, saving candidates 25% application time ### Quible.io — Software Engineer Intern (Nov 2023–Mar 2024, NYC, Remote) - Developed backend microservices using Go and gRPC - Built authentication service with AWS KMS - PostgreSQL with SQLBoiler ORM reaching 150MB/s peak throughput ### UC Berkeley RIOS Lab — Compiler Engineer (Jun 2022–Aug 2022, Berkeley, CA) - Proposed and applied LLVM instruction scheduling optimization for PicoRio 2.0 (RISC-V architecture CPU) - Verified functionality on CPU simulator model and developed unit tests, achieving 40% performance improvement on RISCV-Test benchmark ### Carnegie Mellon University — Research Assistant (Aug 2021–Nov 2021, Pittsburgh, PA) - Developed and optimized a ResNet-based CNN model using Keras for night-time vehicle steering prediction - Achieved 12% average reduction in RMSE, MAE, and MSE; reduced training time by 35% with CUDA-accelerated distributed training - Published: Prediction of Steering Wheel Angle at Night based on CNN, IEEE AIID 2022 --- ## Projects ### ResuMaster AI-powered resume editing platform with 70K+ resumes processed. Chat-based editing with GPT-4o, Chrome extension for job auto-fetching. Tech: React, Flask, OpenAI, Chrome Extension Link: https://www.resumaster.ai/ ### Twitter Streaming Microservices Scalable event-driven microservices using Kafka for Twitter streaming data analysis. Spring Cloud Config for configuration management. Tech: Java, Kafka, Spring Cloud, Docker ### Car Cloud Communication (TLS) Secure web backend in Go with TLS encryption for autonomous driving data. Real-time data capture from Xpeng Motors using ROS and SDK. Tech: Go, Gin, MySQL, Docker, ROS Link: https://github.com/icecreamlun/Construction-of-Car-Cloud-Communication-Website-Based-on-TLS ### Steering Wheel Angle Prediction ResNet-based CNN model to predict vehicle steering angles in nighttime conditions. 12% improvement in RMSE/MAE/MSE. Published at IEEE AIID 2022. Tech: Keras, ResNet, NumPy, Python Link: https://ieeexplore.ieee.org/document/10065435 ### PicoRio RISC-V Scheduling LLVM instruction scheduling optimization for RISC-V architecture CPU. 40% performance improvement on RISCV-Test benchmark. Tech: C++, LLVM, RISC-V, TableGen Link: https://gitlab.com/picorio --- ## Hackathon Awards ### 🥇 1st Place Overall — ByteDance Trae Hackathon (Jan 2026) Built Sentinel QA — an AI agent that automates end-to-end testing using real browser automation. It reads code, generates test plans, opens Chrome, and tests like a real user. Features smart test source detection, automated test planning with Mermaid diagrams, evidence-based reports with screenshots and merge verdicts, and automatic bug fixing with re-verification. Tech: Gemini 2.5 Flash, Puppeteer, Tree-Sitter, Lovable, Render Link: https://shipyardhq.tech/projects/489f5032-a132-49a5-922f-770c4228c659 ### 🥈 2nd Place Overall — MiniMax Hackathon (Feb 2026) Built Axolotl — a VS Code extension powered by multiple AI models for autonomous QA. Features AST analysis via Tree-Sitter WASM across 15+ languages, real browser-based E2E testing with Puppeteer, and a polished React webview UI built with Lovable. Tech: MiniMax, Tree-Sitter, Puppeteer, You.com API, Lovable Link: https://qaxolotl.lovable.app/ ### 🏆 Top 3 Overall ($2,000 Cash) — AWS AI Agents Hackathon (Oct 2025, San Francisco) Built CookieLens, an AI-powered compliance agent that scans any website and auto-generates multilingual cookie policies and compliance analysis aligned with GDPR, CCPA, and SOC2. Selected as Top 3 from 200+ global teams. Tech: AWS Bedrock, Puppeteer, Semgrep, Claude, Vanta ### 🏆 Best Use of LangFlow Award — Hacking Agents NYC Hackathon (May 2025, New York) Designed a multi-agent AI system powering an AI-generated RPG game where players co-create dynamic storylines with autonomous LLM agents. Tech: LangFlow, Multi-Agent AI, LLM, DigitalOcean --- ## Education - M.S. Computer Engineering — New York University, Tandon School of Engineering - Research experience at UC Berkeley RIOS Lab and Carnegie Mellon University --- ## Key Technical Skills AI Engineering, Large Language Models (LLMs), AI Coding Agents, RAG Pipelines, AST-based Code Analysis, Knowledge Graphs, Multi-Agent Systems, LLVM Compiler Optimization, RISC-V Architecture, Go, Python, Java, TypeScript, React, Flask, Spring Cloud, Kafka, Apache Flink, AWS (Bedrock, SageMaker, EC2, KMS), Kubernetes, Docker, PostgreSQL, Redis, Qdrant, Supabase, gRPC, FastAPI