Hann Wang

UCLA

Doctor of Philosophy (PhD), Mechanical and Aerospace Engineering

January 1, 2010 – January 1, 2014

The Johns Hopkins University

MSE, Mechanical Engineering

January 1, 2008 – January 1, 2010

National Taiwan University

BSE, Mechanical Engineering

January 1, 2002 – January 1, 2006

Google

Staff Software Engineer, Office of the CTO

January 1, 2025 – Present

Amazon

Applied Scientist, Amazon AGI

September 1, 2019 – November 1, 2024

Roche

Principal Deep Learning Engineer

July 1, 2017 – August 1, 2019

Genomic Health

Senior Data Scientist, Big Data Analytics

February 1, 2016 – July 1, 2017

Geneio Inc

Cofounder, Lead Data Scientist

November 1, 2014 – February 1, 2016

UCLA Jonsson Comprehensive Cancer Center

Research Scientist

July 1, 2014 – December 1, 2014

Los Angeles Metropolitan Area

Inference of Personalized Therapeutics by Deep Learning Model of Cancer Genomics

January 1, 2015 – Present

- Developed the mathematical model and implemented the Python code of a Hybrid Deep Learning Model which takes advantage of unsupervised pre-training to perform feature extraction from multiple sources, including gene mutations, gene expressions, methylation, chemical fingerprint, drug targets and lineage. The model was implemented with Theano and Pylearn2 to accelerate the code with GPU computing. - Deployed the deep learning code on AWS by Starcluster to facilitate parallel computing, and performed hyperparameter optimization. - Wrote object-relational API, deployed, and managed MongoDB, PostgreSQL databases to integrate multiple genomic and chemical data sources, and to control and record the numerical simulations.

Isoform Inference from Next-generation Sequencing Data

March 1, 2014 – Present

- Wrote and debug the Python code for maximum likelihood inference of noisy sequencing readout. Applied breadth first search to isoform variant graph traversal.

Homologous DNA Sequence Alignment and Inference of Genomic Sequence Variant

October 1, 2013 – November 1, 2013

- Wrote and tested the Python code to implement the dynamic programming algorithm to allow fast alignment of genetic sequences. - Applied Bayesian network inference to detect genetic mutations (SNP) from noisy DNA sequence readout.

Human Facial Image Reconstruction, Recognition, and Object detection

October 1, 2013 – December 1, 2013

- Programmed Eigen face representation of facial images using PCA and reconstructed testing images using them as basis. Applied Fishers discriminant analysis for gender classification. - Developed Matlab model for AdaBoost by constructing weak learners and applied for facial detection algorithm. The result was compared with RealBoost. - Implemented object classification using SVM library for objects such as car in different orientation and pedestrians by training different SVM model and combining them.

Searching of Optimal Combinatorial Cancer Therapy by Hybrid Network and Machine Learning Algorithms

June 1, 2012 – June 1, 2014

- Lead the team of scientists and engineers in this multidisciplinary project involving Machine Learning, High-throughput Robotic Cancer Drug Screening, Database Integration, and Design of Experiment. - Solving the combinatorial problem of multiple drug optimization by developing the mathematical model of a novel machine learning algorithm which combines Biological Network Feature Extraction and Gaussian Process Kernel Regression to generate drug efficacy projection. - Integrated large-scale messy biological dataset from multiple sources to achieve information gain. The dataset was assembled in PostgreSQL from sources including CCLE (cell genomic profile), STRING (protein network), STITCH (drug-chemical interaction), COSMIC(drug screening), all of which were used to train the machine learning model. - Verified the drug efficacy prediction with high-throughput cell-based screening, wrote the Matlab code to generate the input files to control and simulate a pipetting robot’s process.