STAC Summit, 15 May 2019, London

• Dr Michel Debiche, Director of Analytics Research, STAC

Michel will discuss progress toward benchmark standards for ML/DL techniques and technologies based on problems that finance firms in the STAC Benchmark Council have said they care about.

• Dr James Tromans, Gardening Leave
• Dr. Darko Matovski, CEO, causaLens
• Dr John Ashley, Director, Global Financial Services Strategy, NVIDIA [slides]

Financial firms are increasingly using machine learning or deep learning (broadly, “AI”) to improve outcomes or automate functions. Like many other key components of a business plan, AI is a team sport. A brilliant researcher can't be productive without a lot of help with data, technology, and process management. Fortunately, many of these skills exist already in the finance industry, whether it is managing large amounts of data, designing systems for high performance, or managing uncertain processes in an agile way. What are the key engineering and development skills needed on effective data science teams? What technologies are most crucial to understand today and in the near future? How much data science knowledge does a technologist need in order to be a highly effective data engineer or ML engineer? Which parts of the process are most likely to be automated, purchased, or outsourced in the near future, and which parts will continue to rely on in-house (human) technologists? Our panel will debate.

• Dr Alex Allan, CTO & Co-founder, Kortical

Delivering end-to-end AI solutions is a complex, multi-step process. Studies show that many firms struggle with one or more of these steps. The challenges increase in heavily regulated environments like financial services. Kortical is a startup that has built an AI-as-a-service platform from which it serves enterprises in multiple industries. In this talk, Alex will walk through the process of delivering effective AI once data has been prepared—from feature engineering, through model training, tuning, and backtesting, to deployment and life-cycle management of models as APIs—highlighting some of the key challenges and providing tips to overcome them.

• Andrew Jones, VP, HPC Consulting & Services, Numerical Algorithms Group (NAG)

When studying the ROI of new technologies, firms pay great attention to the "R", such as: how much a new architecture will reduce time-to-market for our models/enable us to meet new regulatory requirements/improve our data scientist productivity/etc. But quantifying ROI also requires truly understanding the "I". This may sound simple, but it can be complex for high-performance systems such as HPC grids, machine learning clusters, or big data farms. The total cost of ownership (TCO) of such systems involves many technical and human elements, even if the system is based in a public cloud. At NAG, Andrew spends considerable time advising organizations on how to model the TCO of potential investments in order to guide decisions. In this talk, he will propose a TCO calculator and take audience feedback on it with respect to different use cases important to financial firms.

• Dr Michel Debiche, Director of Analytics Research, STAC

A lot of innovative solutions for big data and big compute workloads like enterprise tick analytics, strategy backtesting, and market risk are being subjected to STAC Benchmarks. If he can catch his breath, Michel will take us through the latest research.

• Richard Croucher, DE & Global Head of Specialised Infrastructure Engineering, Barclays
• Kristie Mann, Director of Product Management, Intel
• Gurpreet Singh, CEO, Pavilion Data Systems
• Derek Burke, Regional Sales Manager, Weka.IO
• Steve Wallo, CTO, Vcinity

Ever-increasing demands from the business, the growth of new AI workloads, and the increasing pull of public clouds call for faster and more flexible ways to get data where it’s needed in computations. Fortunately, the vendor community is competing and collaborating to solve these problems. Innovations include standards like NVMe and NVMe over Fabrics, persistent memory with interesting properties, new and enhanced distributed file systems, rapidly evolving cloud services, and products that claim to substantially mitigate problems we expect when compute and storage are separated by long distances. Sorting through this Cambrian explosion of offerings is a good problem to have, but a problem nonetheless. How should application developers think about memory and storage today? What’s possible with file, block, and object storage? When is public cloud a realistic option, and when not? What are the promise and perils of disaggregated storage? Our panel of experts will debate.

• Peter Lankford, Founder & Director, STAC

Peter will discuss the latest research and Council activities related to low-latency/high-throughput realtime workloads.

• Dan Zheng, CEO, Tick Tock Networks

As financial services firms rapidly increase their use of public clouds for large batch computation, another swathe of use cases—realtime trading—lies in wait. If major markets and their participants could start trading in general purpose public clouds, the promise is both lower cost colocation and the ability to integrate the fast loop of live trading with cloud-scale analysis for faster adaptation of algorithms. But these realtime workflows require low latency, high throughput, and determinism. Some ingredients are emerging in the cloud, such as bare metal instances and realtime market data feeds, but a missing link has been accurate timestamping. Even before regulations like MiFID 2 and the CAT NMS plan began dictating accuracy requirements, trading firms and venues have needed highly accurate timestamps to understand, prove, and improve what happens in their realtime systems. Dan will argue that machine learning-based synchronization algorithms can make such accuracy—even nanosecond accuracy—feasible at cloud-scale. He will also argue that in cases where fairness is a key requirement, such as exchanges or internal matching, the real answer is not deterministic latency but rather highly accurate timestamps that can give commodity, best-efforts networks the effective determinism of a high-performance network.

• Dr Stephen Warr, Head of Technology, XR Trading Europe
• Dr Matthew Grosvenor, VP Technology, Exablaze
• Matt Davey, Director of Product Management, Corvil
• Dr David Snowdon, Director of Engineering, Arista
• Nadeem Zahid, VP Product Management, cPacket

Wire capture is a crucial element of trading systems today, with the resulting data serving as a source not only for operational monitoring but also for the development and backtesting of trading strategies. And like other parts of a trading system, capture architectures can't stand still. In an interactive session with the audience, our panelists will debate key questions like: how are 40G and higher bandwidths affecting requirements? how is the end-to-end workflow evolving, from timestamping and aggregation to capture, replication/distribution, and analysis? how are people balancing speed of capture and speed of retrieval? what impact will recent advances in storage and memory have? what’s changing in the monitoring industry and in the build-vs-buy equation?