In finance, many common calculations are more or less just linear algebra – but at a massive scale and done very fast. Our team developed a portfolio analysis framework for the Aladdin Wealth™ platform, which depends heavily on linear algebra to compute factor risk and achieves significant scale with 15+ million portfolios analyzed daily in batch and 3+ million over API.
However, we've also learned that achieving significant scale and delivering quality products to clients takes more than just efficient code. We've discovered key lessons in our development journey, including the importance of trimming your computational graph, storing data in multiple formats, evaluating technology choices on a total cost basis, leveraging open source, and considering multiple dimensions of modularization. We'll be exploring these topics and other insights in this talk.
Speaker
William Chen
Director and Technical Fellow @BlackRock
Will is a Director and Technical Fellow in the Aladdin Wealth™ business within the Aladdin organization at BlackRock. Aladdin® is BlackRock’s investment and risk management technology platform that unifies the investment management process. Aladdin Wealth™ is BlackRock’s wealth tech business, offering a lineup of technology products serving private banks, retail banks, and independent wealth managers globally. Will’s team is responsible for building the core financial capabilities of the Aladdin Wealth™ product as services and libraries, including the underlying high-scale portfolio risk calculation and optimization engines.
Will personally wrote many of the original components of the Aladdin Wealth™ product as Spring servers and Spark jobs, including the underlying calculation engines analyzing millions of portfolios daily. He is passionate about performance and heap usage, and he was responsible for many of the key scale innovations and optimizations that enable fast and highly available APIs within the Aladdin Wealth™ platform.
Will started his career at BlackRock as an analyst in the Portfolio Analytics Group.