Hi, I'm a machine learning engineer focusing/focused on multiple domains: 3.5 years NLP, 3.5 years recommender-system, 2 years quantitative finance, 0.5 years model-compression.
I have experience in machine learning, backend engineering and data engineering domains, following are the tech stacks I used before:
- Programming Languages: Python, C , Java (Only for Data-Engineering), SQL, JavaScript, Shell, Rust (A Little)
- Frameworks, Libs or Tools: Pulsar, Milvus, gRPC, Spark, Hive, K8S, PyTorch, FAISS, Redis, Flask, TensorFlow (Long Time Ago)
Here is my projects index:
-
Side Projects
- simpler-distil-whisper: Reproduce paper Distil-Whisper: Robust Knowledge Distillation via Large-Scale Pseudo Labelling.
- summary-evaluator: Reproduce paper BERTScore: Evaluating Text Generation with BERT.
- PLM-ICD-multi-label-classifier: Reproduce paper PLM-ICD: Automatic ICD Coding with Pretrained Language Models.
- feather: A C feature-hash lib with Python binding provided.
- osimhash: A Python binder over simhash C text deduplication lib.
- pypack: Generates Python runtime tar.gz file (for PySpark) runnable on all python-version/os/platforms.
-
Codes Reading
- fastTextAnnotation: The very detailed code annotation for facebook fasttext lib.
- hnswlibAnnotation: The very detailed code annotation for hnswlib.
- finBERT: BERT for financial news sentiment classification.
-
Self Using
- quicmd: Some useful quickly execution commands.
- config4: Some self-using configs, for now about tmux and vim.
- wiki4codes: Some lib/framework/algorithms/models' trials recording, demos, examples, etc...