James Chang

📝 About Me

About me: I'm James (Hung-Yang) Chang, an AI Software Developer at Cerence, specializing in LLM function calls, edge AI, and hardware acceleration. My work focuses on optimizing AI systems for automotive applications and edge devices, combining software engineering with machine learning expertise.

🎓 Education

B.S. in Electrical Engineering

National Tsing Hua University, Hsinchu (NTHU), Taiwan

Sep. 2014 - Jun. 2018

Overall GPA: 3.99/4.3 (3.87/4)

MSc (thesis) in Electrical & Computer Engineering

McGill University, Canada

Sep. 2020 - Feb. 2023

Overall GPA: 4/4

💼 Experience

AI Software Developer, Cerence

Aug. 2023 - Present

Machine Learning Engineer, Bittensor, Opentensor

Mar. 2023 - Jul. 2023

Graduate Research Assistant, McGill Edge Intelligence Lab

Sep. 2020 - Feb. 2023

Research Intern, IBM Neuromorphic Devices and Architectures

Oct. 2018 - Apr. 2019

Research Assistant, NCTU DSML Group

Oct. 2017 - Oct. 2018

Undergraduate Researcher, NTHU SISAL

Feb. 2017 - Feb. 2018

📚 Selected Publications

📄 PipeBERT: High-throughput BERT Inference for ARM Big.LITTLE Multi-core Processors

Journal of Signal Processing Systems, IEEE SIPS 2022

Hung-Yang Chang, Seyyed Mozafari, Cheng Chen, James Clark, Brett Meyer, and Warren Gross

📄 AI hardware acceleration with analog memory: micro-architectures for low energy at high speed

IBM Journal of Research and Development

Hung-Yang Chang and Geoffrey W. Burr, Pritish Narayanan, Stefano Ambrogio, et al.

📄 High-Throughput Edge Inference for BERT Models via Neural Architecture Search and Pipeline

GLSVLSI 2023 (Poster Presentation)

Hung-Yang Chang, Seyyed Mozafari, James Clark, Brett Meyer, and Warren Gross

📄 A Novel Architecture to Build Ideal-linearity Neuromorphic Synapses on a Pure Logic FinFET Platform Featuring 2.5ns PGM-time and 10^12 Endurance

2019 Symposium on VLSI Technology (Oral Presentation)

E.R Hsieh, H. Y. Chang, Steve S. Chung, S. Simon Wong et al.

🚀 Projects

Hardware Aware Efficient Training Competition (ICLR 2021 Workshop) [GitHub]

Feb. 2021 - Mar. 2021, ECSE, McGill

Built a modified ResNet with mixed-precision training to achieve 95%, 77% validation accuracy for CIFAR-10 and CIFAR-100, respectively in 5 minutes with single V100 GPU

Exploring Super-Converge in Analog NNs with IBM tool (Deep Learning (ECSE 552)) [GitHub]

Feb. 2021 - Apr. 2021, ECSE, McGill

Explored super-convergence phenomena in IBMs Analog Hardware Acceleration Kit for in-memory training of DNNs. Applied cyclic learning rates to VGG8, ResNet18, and LeNet architectures on MNIST and CIFAR10

System C implementation: Design of MPSoC (Design of Multiprocessor System-on-chip (ECSE 541)) [GitHub]

Oct. 2020 - Dec. 2020, ECSE, McGill

Utilized IBMs Analog Hardware Acceleration Kit and PyTorch to simulate in-memory FCNN & CNN computations for MNIST Dataset

📱 App Gallery

🐕

Puppy Step

Companion app for tracking pet growth and milestones with real-time tracking and milestone gallery

Live Demo

💳

Subscribe Manager

Track recurring expenses and manage subscriptions efficiently with intelligent insights

Preview

🇨🇦

Canada Count Down

Track timeline from Permanent Residency to Canadian Citizenship with precision

Open App

📋

Coach Book IQ

Smart team management and scheduling system for professional coaching staff

Launch App

⚡ Skills

💻 Programming Languages

Python, Pytorch, Pytest, Tensorflow, TVM, Matlab, SystemC, LATEX

🔧 Engineering Tools

DevOps, gPRC, Iceberg Catalog, MCP, Google A2A, Flask, Docker, Kubernetes, Jira

🏆 Awards & Honors

Graduate Excellence Fellowship - Awarded with 4600 CAD for 10 selected graduated students (ECSE, McGill, 2022)
Outstanding Project Award in Contest of Implementation - Top 10 of Research project competition with more than 250 student competitors (EECS, NTHU, 2018)
International Volunteer Certification - Awarded with $1000 USD funding to host 100 people classes, and school anniversary fair in Malaysia (Ministry of Education Taiwan, 2015)

🏠 Hobbies/Personal

🏃‍♂️ Sports & Fitness

🏃

Running

Regular runner with a personal best of 1h47m in half marathon

🏊

Swimming

Regular swimmer with 26min time for 1000m distance

🏀

Basketball

Enjoy playing basketball recreationally and following the sport

💪

Gym & Fitness

Consistent gym routine focusing on strength training and overall fitness

⛷️

Skiing

Current at Blue Square level, aiming toward Black Diamond runs

🏃

Strava

Track my running and fitness activities 🔗 Follow on Strava

🐾 My Pets 🐾

Nana 🐕

Cocker Spaniel who loves playing fetch, has amazing energy, and enjoys running

Nini 🐕

Poodle who loves to cuddle and be affectionate like a little furry caterpillar

Poyno 🐈

Mischievous cat who loves to play with Nana and jump around everywhere