Liu Jia
Backend and machine-learning engineer with internships at Tencent and ByteDance, building high-concurrency systems and recommendation models. I like owning a problem end to end — from distributed-systems design to shipping measurable gains in production — and am now seeking an MSc to deepen my research in distributed systems and applied ML.
Education
Sun Yat-sen University
GuangzhouGPA 3.8 / 4.0, Top 5% of major
Relevant Coursework: Data Structures & Algorithms, Machine Learning, Operating Systems, Database Systems, Computer Networks
Professional Experience
Tencent
Shenzhen- Rearchitected the WeChat Pay reconciliation pipeline, cutting the daily batch runtime from 4 hours to 45 minutes.
- Built a high-concurrency reconciliation service in Go, handling ~20M transactions per day.
- Raised core-module test coverage from 62% to 91% with unit and integration tests.
ByteDance
Beijing- Optimized recall models on the recommendation team, improving offline AUC by 1.8%.
- Reproduced and improved a two-tower recall model in PyTorch, lifting CTR by 3.2% after launch.
Selected Projects
- Implemented a strongly-consistent distributed key-value store on the Raft consensus algorithm with dynamic membership.
- Built an LSM-Tree storage engine reaching 120K write ops/s.
Tech: Go, Raft, LSM-Tree, gRPC
Chinese Medical QA
- Built a retrieval-augmented medical Q&A system over 300K structured knowledge entries.
- Improved answer accuracy from 71% to 86% on an in-house test set.
Tech: Python, RAG, LLM, FAISS
Awards & Honors
Silver Medal, ICPC Asia Regional Contest
National Scholarship
Awarded to the top 1% of students by overall ranking.
Technical Profile
Languages
English: IELTS 7.5 (Listening 8.0)
Mandarin: Native