ABOUT ME
Meng Chen
M.S. Computer Science @ UNC-Chapel Hill•AI agents & software engineering
I am an M.S. Computer Science student at UNC-Chapel Hill, advised by Huaxiu Yao. My research focuses on evaluating self-improving AI agents and running reproducible harness-optimization experiments. I also independently build and maintain Serenity Intelligence / Alphaline, an AI information platform that has served 100+ paying customers to date. I am seeking Summer 2027 internships in software engineering and applied AI.
NowChapel Hill, NC
Evaluating AI agents and building software for real users.
CURRENT FOCUS
AVAILABLE FOR
Email: mengc@cs.unc.edu
Selected Work
Projects
An independently developed product with paying customers and an end-to-end software stack.
Independent Developer | May 2026 - Present
Serenity Intelligence / Alphaline
An AI information and subscription platform for technology and investment content. Built and maintained for 100+ paying customers to date, with 6,800+ posts ingested and 6,700+ structured analyses stored.
- • Streaming ingestion with heartbeat monitoring and backoff reconnection, scheduled polling, and content deduplication.
- • Multiple LLM providers with first-success selection, schema validation, JSON repair, retries, and original-content fallback.
- • Channel-level access controls, subscription-expiry management, notification workflows, and an interactive stock-information dashboard.
Toolkit
Skills
Tools I use to build applications and run reproducible AI experiments.
Languages
Programming and data analysis.
Web & Data
Applications, APIs, and persistent data.
Research & Tools
Reproducible experiments and agent evaluation.
Work
Research
Two ongoing projects at UNC-Chapel Hill, advised by Huaxiu Yao.
RSI-Exam: Self-Improving AI Agent BenchmarksAug 2026 - Present
UNC-Chapel Hill | Advisor: Huaxiu Yao
- Contribute to scientific ML benchmark design, defining quantitative objectives, optimization spaces, and held-out evaluation protocols. Investigate information leakage and reproducibility.
- Develop and evaluate an enhancer-gene prediction task using CRISPR labels and genomic features, with chromosome-based splits and AUPRC evaluation. Analyze a 27-node agent experimentation trajectory.
- Run a Python-based multi-agent review workflow across approximately 30 iterations, debugging structured-output parsing, persistent state, and iteration continuation.
LLM Agent Evaluation & Harness OptimizationAug 2026 - Present
UNC-Chapel Hill | Advisor: Huaxiu Yao
- Configure Harbor-based environments on shared Linux servers and run benchmark and harness-search workflows, preserving evaluation jobs, candidate artifacts, and provenance.
- Debug Harbor/LiteLLM integration with an OpenAI-compatible gateway, resolving provider-routing and API model-name mismatches before launching experiments.
- My contribution centers on experimental setup, execution, and validation within the broader harness-optimization research project.
Selected Publication
AutoResearchClaw
Co-author · arXiv preprint · 2026
Self-Reinforcing Autonomous Research with Human-AI Collaboration
A multi-agent research system combining iterative experimentation, verifiable reporting, and human-AI collaboration.
Background
Education
Computer science graduate study supported by a foundation in scientific research.
University of North Carolina at Chapel Hill
Aug 2026 - Expected 2028
M.S. in Computer Science
Advisor: Huaxiu Yao. Current research: AI agent evaluation, scientific ML benchmarks, and agent harness optimization.
Shanghai Jiao Tong University
Sep 2021 - Jun 2025
B.S. in Medical Laboratory Technology
Recipient of the Bo Le Undergraduate Research Funding award: one of 30 recipients, with RMB 100,000 in research support.
Coursework: Data Structures in C++, Linear Algebra (A), Discrete Mathematics (A), Computer Architecture (in progress), Calculus, Python Programming, C Programming, Biostatistics, and Bioinformatics (A).
Let’s build
Contact
I am seeking Summer 2027 internships in software engineering and applied AI.
Let’s talk about software, AI agents, and research.
Chapel Hill, NC