Associate (Developer Machine Learning) Full-time

About Polymer

A market-neutral, multi-manager platform investing in liquid securities with a strong Asia focus, Polymer Capital combines established institutional support and deep knowledge of local financial markets with a dedication to discovering and developing the region’s best investment talent. Polymer has six offices around Asia-Pacific.

Job Description

As a software engineer, you will work closely with ML researchers and other developers in the team.

Our team develops machine learning/deep learning models for quant strategies.

We’re looking for machine learning engineers to design, develop and maintain systems that assist ML researchers with various tasks including storage, data ETL, model training, validation and hyper-parameters tuning.

Our team follows Agile & DevOps practices, developers are expected to handle solutions end-to-end.

• Based on data types, access patterns, design systems including but not limited to:

1. Data storage
2. Data ETL
3. Workflow

• Integrate ML models for offline and online inference
• Help researchers with technical issues, best practices and performance tuning

Requirements

• Bachelor or Master’s degree in STEM majors
• 2 – 3 years of ML project experience
• 1 – 2 years back-end experience using languages like C#/Java, C++/Rust, Python
• Familiar with Machine Learning related frameworks, such as TensorFlow and PyTorch
• Experience managing infrastructure, comfortable with Linux
• Willing to step out of comfort zone and learn, good at problem-solving
• Good theoretical foundation of machine learning and data mining
• Able to read and write in English, limited speaking capacity is acceptable

Good to Have:

• Well versed in AWS / GCP
• Understand different data storage solutions, their PROs and CONs
• Good mathematics and statistic knowledge (understand concepts like matrix and related operations, correlation, standard deviation, etc.)
• Experienced in automated hyperparameter tuning optimization, automated model training and scoring and model selection