Internship & project matching
Yanxuan
Students find internships and projects that fit them, companies post roles to specific schools, and the school sets the rules, reviews results and keeps the record. Project and internship matching share one workflow.
Illustrative interface with sample data
The student view above is live: search, filter, save and apply. Applications appear on the board below.
Reasons, not rankings
Each recommendation says how it relates to the student's courses, exercises and interests.
Requirements up front
Missing prerequisites or schedule conflicts are flagged before anyone applies.
The school oversees it all
Configurable rules, with review, publication and export all on record.
See where every application stands.
From submission and school review to interviews, offers and the internship itself, each card says what's happening, who's handling it and what comes next.
The school sets the rules. The AI drafts the match.
Both sides confirm.
Supervisors publish projects and places; students rank their choices. Press “生成推荐” (generate): the system proposes an allocation within the constraints, with reasons, and lists conflicts and open places.
课题与名额
6 个课题 · 14 个名额 · 28 名学生已填报
| 课题 | 名额 | 报名 | 推荐 / 名额 | 状态 |
|---|---|---|---|---|
| 低功耗 SRAM 设计陈老师 | 2 | 7 | 待匹配 | |
| 片上网络路由器王老师 | 3 | 4 | 待匹配 | |
| RISC-V 流水线优化李老师 | 2 | 9 | 待匹配 | |
| 光电探测器建模林老师 | 1 | 0 | 待匹配 | |
| 电源管理芯片赵老师 | 3 | 3 | 待匹配 | |
| FPGA 图像加速孙老师 | 3 | 5 | 待匹配 |
- !RISC-V 流水线优化:9 人报名,2 个名额
7 人按第 2 志愿进入调剂;候补顺序已生成,供李老师参考。
- 0光电探测器建模:暂无人报名
建议向本学期《光电子学》成绩前 30% 的学生定向推送,并请林老师补充课题描述。
Illustrative interface with sample data
Describe the role in a paragraph.
The AI writes the posting.
Companies describe what they need in plain language; the AI structures field, skills, requirements and timing, and flags what's missing. After the company confirms, it goes to the chosen schools and is visible once career staff approve.
- 未写明实习补贴,学生普遍关注,建议补充。
Rules for when
things go wrong.
Matching is only trusted if edge cases are handled openly. Each case has a defined rule, a full record, and a person who decides.
Oversubscribed
Ranked by the school's rules; students not placed move to reallocation, with reasons.
No applicants
The supervisor is prompted to improve the description, or it's shown to students from related courses.
Supervisor declines
The place is released and the waitlist moves up, with notifications.
Student withdraws
The place returns to the pool; the record is kept.
Conflicting picks
If a student is recommended for several projects, only the higher preference stays, with the reason.
Final review
Appeals during the publication period; once locked, results are exported and archived.
During the pilot we'll compare manual allocation, preference-only allocation and system recommendations on the same data, measuring match rate, preference satisfaction, capacity use and reallocations. Results will be published after validation.
Other products
Smart Teaching
In developmentA practice-first digital TA. It runs Verilog simulations, hints before it explains, and reports to the teacher's dashboard.
SwiftLearn
In internal useTurns a lecture into text as it happens, then organises it into searchable notes and key points.
PaperTree
PlannedPaper sharing and a knowledge tree for research groups, so a lab's reading outlives each graduating class.
Start with one real workflow.
Tell us what you want to solve and we'll work out where to begin.