• Bachelor of Engineering (Information Engineering and Media)
• Honours (Distinction); CGPA: 4.47 / 5.00
• MOE SM2 Scholarship Recipient

• Aalto University, Finland Jan 2022  -  Jun  2022
• National University of Singapore Jan 2021  - May 2021

• Built a kubectl-style Python CLI to provide a unified interface for Illumina instrument management, improving developer experience and reducing test setup time by 30%.
• Overhauled idempotent Bash lifecycle tooling for Kubernetes-based DNA sequencing software, enabling one-click installation in-field deployment, cutting deployment incidents by 25%.
• Built HTTP server to expose stateful AI agent workflows as endpoints, enabling zero-cost org-wide AI adoption with automated test failure triage, server debug, and PR review; cut on-call and triage hours by 40%.
• Built a self-improving agent system with evals, which self-analyzes session transcripts to autonomously improve knowledge base, memories, and skills; increased agent bug-fix PR acceptance rate from 70% to 93%
• Forked and improved a Jenkins CLI to be agent-first, closing the self-verification gap in agentic workflows, enabling fully autonomous agent to complete “test → triage → fix → test” long running tasks.
• Implemented Andrej Karpathy's auto-research on test optimization, reducing PR CI run time from 80s to 10s, and nightly e2e suite from 4hr to 2hr 37min.

• Responsible for quality assurance for LiveNet CDN live streaming infrastructure, and BytePlus Live SaaS across server, web and mobile; supporting livestreams with over 500K PV on BytePlus and TikTok.
• Optimized and deployed PyTest automation in CI/CD among cross-region data centers; shortened release cycles by 30% while maintaining over 99% pipeline stability.
• Initiated team transition from providing test support to delivering testing capabilities; built QA no-code platform, and developed CLI for API robustness test generation, increasing developer self-testing rates by 20%.

• Conducted extensive image data collection, pre-processing and dataset generation with Pandas and OpenCV in Python; created consistent data flow for 5 computer vision Machine Learning Models.
• Utilized TensorFlow and PyTorch to finetune and evaluate 2 Semantic Segmentation and 3 Object Detection models on Linux; achieved 94.4% accuracy with YOLO Algorithm.
• Continuously monitored model performance, implemented batch processing in real-world environment for real-time object detection, which improved response time by 15%.

• Open Source (github.com/baggiiiie): maintainer of jjui (2K stars), the most popular TUI for jj-vcs.
• SaaS: created simple-cv.app, reached 200 monthly active users.
• Writings: writing technical blogs at yingchao.dev/blog, covering topics like agent memory and prompt caching.