I build the systems that turn foundation models into products.

Software engineer at Meta working on AI infrastructure, AI products and ecosystem, and ML foundation models. I work across the stack that sits between a model and a user: training and serving infrastructure, agent and retrieval pipelines, evaluation, and the product surfaces where model behavior finally meets people.

Now
AI Engineer, Meta
Work
AI Infra · AI Product & Ecosystem · Foundation models
Research
LLM systems, multi‑agent tooling, applied ML

About

At Meta my work spans AI infrastructure, AI product and ecosystem systems, and ML foundation‑model initiatives: the pipelines that train, serve, and evaluate models, and the product systems that put them in front of people. I care about the whole loop, from model capability to what a user actually experiences.

I came to this through research and still publish. My recent work covers multi‑agent systems for code review and literature synthesis, LLM evaluation and alignment with human judgment, and applied ML for clinical prediction. Before Meta I built data platform infrastructure at Meltwater and did research on generative AI, NLP, and health ML at Northeastern.

LLM systemsMulti‑agent pipelinesRAG & retrieval Model evaluationML infrastructurePyTorch PythonC++TypeScript FastAPINext.jsKubernetes

Projects

  1. ResearchPilot

    live demo

    A research copilot that retrieves papers, extracts structured findings, synthesizes literature, and drafts citation‑aware related work through a streaming workflow.

    FastAPI · Next.js · DSPy · Qdrant · SSE

    Code Demo Paper

  2. RepoReviewer

    live demo

    A local‑first, multi‑agent code review system for public GitHub repositories and pull requests, with a shared engine behind both CLI and web interfaces.

    Python · FastAPI · LangGraph · Next.js

    Code Demo Paper

  3. CLARA

    A medical dialogue simulation prototype for history taking and diagnostic reasoning, with deterministic and LLM‑enhanced analysis paths and graceful fallback between them.

    Python · OpenRouter · NLP · Simulation

    Code

Publications

  1. 2026

  2. Visualizing High‑Dimensional Graph Embeddings via Informed Multi‑View Projections

    Ya Ji, Xuefeng Li, Timo Brand, Jacob Miller, Peng Zhang, Stephen Kobourov, Yifan Hu

    IEEE VIS · 2606.31119

    arXiv PDF Scholar

  3. Small Language Models for Early Alzheimer's Risk Detection: Accuracy‑Efficiency Trade‑offs in the DMV Framework

    Vinatha Viswanathan, Peng Zhang, Divya Chaudhary

    ARIAL Workshop at IJCAI

    Scholar

  4. SAFE: Segment‑Aware Filtering and Evaluation for Lyric Content Moderation

    Peng Zhang, Jiawen Xie, Zihan Su, Shanu Sushmita

    Culture × AI Workshop at ICML

    Scholar

  5. Beauty in the Eye of AI: Aligning LLMs and Vision Models with Human Aesthetics in Network Visualization

    Peng Zhang, Xuefeng Li, Xiaoqi Wang, Han‑Wei Shen, Yifan Hu

    EuroVis · 2604.03417

    arXiv PDF Scholar

  6. RepoReviewer: A Local‑First Multi‑Agent Architecture for Repository‑Level Code Review

    Peng Zhang

    arXiv preprint · 2603.16107

    PDF arXiv Code Demo

  7. ResearchPilot: A Local‑First Multi‑Agent System for Literature Synthesis and Related Work Drafting

    Peng Zhang

    arXiv preprint · 2603.14629

    PDF arXiv Code Demo

  8. 2025

  9. Hybrid Deep Learning Framework for Enhanced Melanoma Detection

    Peng Zhang, Divya Chaudhary

    IEEE Transactions on Computational Biology and Bioinformatics

    Paper Scholar

  10. Leveraging Clinical Record Geolocation for Improved Alzheimer's Disease Diagnosis Using DMV Framework

    Peng Zhang, Divya Chaudhary

    Biomedicines

    Paper Scholar

  11. Beyond Detection: Predicting Code‑Switch Points in Multilingual Conversations

    Jiawen Xie, Peng Zhang, Jayash Koshal, Shanu Sushmita

    WiML Workshop at NeurIPS

    Paper Scholar

  12. When Pain Hides in Silence: ML‑Driven Flare‑Up Prediction for Hidradenitis Suppurativa Using Synthetic Patient Data

    Shagun Saboo, Peng Zhang, Rishabh Jain, Adwita Arora, Krish Chopra, Sandeep Kumar, Divya Chaudhary

    GenAI4HealthInfo Workshop at IJCAI

    Article Scholar

  13. Leveraging Geolocation in Clinical Records to Improve Alzheimer's Disease Diagnosis Using DMV Framework

    Peng Zhang, Divya Chaudhary

    W3PHIAI Workshop at AAAI · 2502.04288

    arXiv Scholar

  14. 2024

  15. Hybrid Deep Learning Framework for Enhanced Melanoma Detection

    Peng Zhang, Divya Chaudhary

    BIOKDD Workshop at KDD

    Paper Scholar

Experience

  1. Software Engineer, AI Meta

    AI infrastructure, AI product and ecosystem systems, and ML foundation‑model initiatives.

  2. Research Assistant Northeastern University

    Applied machine learning and research engineering, advised by Yifan Hu.

  3. Research Assistant Generative AI Research Lab, Northeastern

    Generative AI, NLP, and multilingual systems, advised by Shanu Sushmita.

  4. Teaching Assistant Northeastern University

    CS6240 Large‑Scale Parallel Data Processing · CS6140 Machine Learning.

  5. Student Chair AAAI 2025 Workshop on LLMs and Generative AI for Health

    Coordinated workshop operations for a research venue on LLMs and generative AI.

  6. Peer Mentor Mentor Collective

    Mentored students in the MS in Computer Science Align program.

  7. Software Engineer Meltwater

    Dashboard platform infrastructure and data pipeline systems.

  8. Founder, SCOPE Camp JA Worldwide, Shanghai

Education

  1. Northeastern University logo

    Northeastern University Khoury College of Computer Sciences

    MS in Computer Science

  2. Shanghai University logo

    Shanghai University

    BS in Business and Information Systems