CV

Personal CV, updated on 15 Jun, 2024

Basics

Name Liangliang Zheng
Label Software Engineer
Email zhengliangliang1997@gmail.com
Phone +32/484131998
Url https://paranoiarchive.com/

Work

  • 2022.11 - Present
    Software Engineer
    ING Amsterdam, the Netherlands
    Developed an alert generation system for an AML orchestration tool. Implemented the front-end using Jupyter Hub. Deployed and maintained different alert definitions, ensuring compatibility with different alert definition versions. Developed automatic testing framework, including Azure test plans, functional tests, regression tests. Assisted the backtesting and look-back team in investigating alerts.
    • Developed an alert generation system for an AML orchestration tool.
    • Implemented the front-end using Jupyter Hub.
    • Deployed and maintained different alert definitions, ensuring compatibility with different alert definition versions.
    • Developed automatic testing framework, including azure test plans, functional tests, regression tests.
    • Assisted the backtesting and look-back team in investigating alerts.
  • 2020.10 - 2022.10
    Data Scientist
    Euroclear Brussels, Belgium
    Reduced manual labor by 60 man-days daily by developing prospectus extraction models. Fetched 6 months of training label data from datalake and combined with text data queried from Hbase. Built (22/59) models (Categories: Classification, Extraction, Rule based) to extract general, final redemption, coupon fields in the prospectus. Integrated evaluation heatmap and report generation functions allowing developers and business side to effortless build performance tracking dashboard.
    • Reduced manual labor by 60 man-days daily by developing prospectus extraction models.
    • Fetched 6 months of training label data from datalake and combined with text data queried from Hbase.
    • Built (22/59) models (Categories: Classification, Extraction, Rule based) to extract general, final redemption, coupon fields in the prospectus.
    • Integrated evaluation heatmap and report generation functions allowing developers and business side to effortless build performance tracking dashboard.

Education

  • 2018.09 - 2020.07

    Brussels, Belgium

    MSc
    Vrije Universiteit Brussel
    Applied Sciences and Engineering: Computer Science (AI)
  • 2014.09 - 2018.06

    Hunan, China

    BSc
    Hunan Agricultural University
    Information and Computing Science

Skills

Programming
Python
SQL
Bash
Programming
Go
C++
Java
Matlab
R
HTML/CSS
JavaScript
Dev & ML
Vim
Git
Docker
Spark
Azure
Cloud Platform
TensorFlow
PyTorch
PowerBI
Other
Linux
Unix
Windows
Excel
LaTeX
Agile Terminology

Languages

Mandarin Chinese
Native speaker
Cantonese Chinese
Native speaker
English
Fluent
French
Basic

Projects

  • 2023.06 - 2024.03
    OLE (Orange Language Engine)
    Engaged in data synthesis, SQL generation DIY projects. Built the end-to-end pipeline for data transformation, data synthesis, and data evaluation.
    • Engaged in data synthesis, SQL generation DIY projects.
    • Built the end-to-end pipeline for data transformation, data synthesis, and data evaluation.
  • 2020.12 - 2021.06
    New Issue Prospectus Extraction
    Reduced manual labor by 60 man-days daily by developing prospectus extraction models. Fetched 6 months of training label data from datalake and combined with text data queried from Hbase. Built (22/59) models (Categories: Classification, Extraction, Rule based) to extract general, final redemption, coupon fields in the prospectus. Integrated evaluation heatmap and report generation functions allowing developers and business side to effortless build performance tracking dashboard.
    • Reduced manual labor by 60 man-days daily by developing prospectus extraction models.
    • Fetched 6 months of training label data from datalake and combined with text data queried from Hbase.
    • Built (22/59) models (Categories: Classification, Extraction, Rule based) to extract general, final redemption, coupon fields in the prospectus.
    • Integrated evaluation heatmap and report generation functions allowing developers and business side to effortless build performance tracking dashboard.
  • 2022.08 - 2022.10
    Anti Money Laundering Graph(AML Graph)
    Built visualization tool for compliance team to visualize and analyze transaction behavior. Built entity resolution pipeline and resolved around 35.3% of all the transaction entities extracted from data lake, write resolved ids back to data lake. Combined resolved ids information and transaction type to build graph nodes and edges, visualizing using pyviz Network.
    • Built entity resolution pipeline and resolved around 35.3% of all the transaction entities extracted from data lake, write resolved ids back to data lake.
    • Combined resolved ids information and transaction type to build graph nodes and edges, visualizing using pyviz Network.
  • 2022.08 - 2022.10
    Transaction Monitoring Re-Calibration
    Improved internal compliance engine by reducing 9.2% of false positive alerts through dynamically re-calibrating alert threshold. Ingested 13 weeks of data to HDFS and replicated the alert detection logic same as in the internal rule-based compliance AML engine. Re-calibrated threshold for different segments and risk levels based on historical threshold percentile distribution.
    • Improved internal compliance engine by reducing 9.2% of false positive alerts through dynamically re-calibrating alert threshold.
    • Ingested 13 weeks of data to HDFS and replicated the alert detection logic same as in the internal rule-based compliance AML engine.
    • Re-calibrated threshold for different segments and risk levels based on historical threshold percentile distribution.