qingqi1106
全职 · 1000/日  ·  21750/月
工作时间: 工作地点: 远程
服务企业: 0家累计提交: 0工时
聊一聊

APP聊一聊

个人介绍

Qingqi Zhang – AI engineer, full-stack problem-solver, and incoming Harvard Applied Math PhD


Hands-on AI product builder. Since Apr 2025 I’ve been founding and coding an AI-driven study-abroad platform that blends a web portal with an LLM-powered chat community, delivering real-time, personalized application advice.Proven programming depth. Gold-medal ICPC alum; fluent in Python, C++, Java, and modern web stacks.Research-grade ML expertise. First-author of Schur Net (NeurIPS 2024) and co-author of an ICML 2024 paper on graph VAEs; built CUDA-level optimizations and profiling pipelines for higher-order GNNs.Industry-level impact. At IBM Research, accelerated graph-explanation algorithms by 1000× while improving interpretability for node-classification models.Future-focused. Starting Fall 2025, I’ll pursue a PhD in Applied Math at Harvard, researching AI for climate modeling—strengthening my ability to architect scalable, mission-critical AI systems.

工作经历

  • 2025-04-11 -至今AI留学创业团队发起人

    我们旨在做一个AI为基础的留学申请平台。通过网站实现从选校、文书、套瓷等的全过程管理,数据库接入AI,可以提供实时、定制化的建议,同时有一个聊天社群鼓励申请者、学长学姐、导师的相互交流。

  • 2021-07-01 -2022-05-31IBM ResearchResearch Affiliate

    As a Research Affiliate at IBM Research (MIT-IBM Watson AI Lab, Jul 2021 – May 2022) I spearheaded explainability advances for graph neural networks. I extended the Contrastive Explanation Method to graphs (CEM-G), relaxing the original 0-1 search to a continuous, sparsity-promoting formulation and devising a refined PP/PN feature-selection criterion for imbalanced graphs. My work produced detailed node-level explanations that revealed three root causes of GNN errors. I also created CF-PGExplain

教育经历

  • 2025-09-01 - 哈佛大学应用数学博士

  • 2022-09-01 - 2024-12-13芝加哥大学计算与应用数学硕士

  • 2018-09-01 - 2022-07-01浙江大学计算机科学与技术本科已认证

语言

普通话母语水平
英语母语水平
0
1
2
3
4
5
0
1
2
3
4
5

技能

C++精通
Java精通
Torch精通
0
1
2
3
4
5
0
1
2
3
4
5
更新于: 2025-05-01 浏览: 13