Build end-to-end cloud-based auto training and serving pipelines on Google Cloud Platform.
• Create standardized feature engineering pipelines using Beam, DataFlow, and Airflow.
• Enable fast data collection, model building, and evaluation via TensorFlow Extended (TFX).
• Lead pipeline standardization with templating and Jenkins CICD
• Provide model explainability with Self-Attention Block

Google Cloud Platform TensorFlow Extended(TFX)
• Beam, DataFlow, Airflow
•
•

Develop decision-making algorithms to detect fraudulent traic in mobile ads and payment.
• Perform big data ETL Through BigQuery and Apache Spark.
• Design scalable ensemble models consisting of self-clustering and supervised learning.
• Explore model training in distributed systems for DNN and Gradient Boosted Tree.

Google BigQuery Apache Spark ETL
•

Design reinforcement learning (RL) environment for unified ads auction
Build deep policy gradient networks (DPGN) to learn auction strategies
Develop light-weighted scalable multi-armed bandit (MAB) methods to optimize revenue
PUBLICATIONS & PRESENTATIONS
2022.05 Graph+AI Breakout: Graph-based Fraud Detection in Unity
Graph+AI Summit 2022
An oral presentation covering how the Trust & Safety team in Unity utilizes graph-based
approaches to facilitate fraud detection.
2021.04 Generating multi-type sequences of temporal events to improve
fraud detection in game advertising
ECML PKDD 2021
A conference paper presenting our study that proposes using a variant of Time-LSTM cells
in combination with a modified version of Sequence Generative Adversarial Networks
(SeqGAN)to generate artificial sequences to mimic the fraudulent user patterns in ad traic. /
/ :()+1 510-277-2731/ ( ): +86 18501302672
| | : jiangl.science@outlook.com
linkedin.com/in/lun-jiang-962838149/ lunj12.github.io
2017 2018
-
3.90/4.0
2016 2017
-
3.88/4.0 : A Parallel GPU Implementation of Smoothed Particle Hydrodynamics
2012 2016
-
89.45/100 2016 (2014 & 2015)
2022.11 Present
-
Unity Technologies
• Unity
• Transformer LSTM
•
• GNN embedding

(DPGN)
multi-armed bandit (MAB)
2022.05 Graph+AI Breakout: Graph-based Fraud Detection in Unity
Graph+AI Summit 2022
Oral presentation: Unity (Trust & Safety)
Graph embedding
2021.04 Generating multi-type sequences of temporal events to improve fraud
detection in game advertising
ECML PKDD 2021
Conference paper: Time-LSTM cells ——
Sequence Generative Adversarial Networks (SeqGAN)
SELF INTRO
Senior Data Scientist and Machine
Learning engineers specialized in Anti-
Fraud and Game Monetization.
Always prepared for new opportunities
and challenges!