Internship
Ant Financial Services Group (Ant Group)
- Researched clustering-based general compression methods for semi-structured data such as logs and JSON, and explored the application of learned index in key-value index scenarios to reduce storage costs.
- Designed a simple compression framework based on clustering results and implemented a compression evaluation algorithm to measure the compression ratio on different datasets; later focused on coding algorithm implementation.
- Internship lasted until the end of 2022, with outcomes planned for publication in research papers.
Competitions
CCF Big Data & Computing Intelligence Contest (BDCI)
ยท National 7th Place (7/295)
- Designed a transaction cycle detection scheme under specific constraints, targeting anti-money laundering and anti-fraud scenarios.
- Pre-computed candidate sets for each edge to reduce redundant checks during queries.
- Adopted Forward Star data structure and multi-threading to accelerate query performance.
- Proposed a bi-directional cycle query method based on DFS: for each vertex, only half a cycle needed to be discovered in each direction, significantly improving efficiency.