Migration of a legacy risk valuation system.
A monolithic legacy system implemented in C++ had to be splitted into smaller services to simplify support and development of new features.

New implementation was mostly (99%) Java-based.
The remaining part -- DBA (Deutsche Bank Analytics) -- a single-threaded library was implemented in C++.

New system was implemented as a set of services, offered to each other and various internal clients (different DB projects).

Entire functionality has been split into products and migrated on a product-by-product basis.

As a lead developer on the project: 

• participated in new system design
• implemented an initial version of Risk Calculator service
• driving further service evolution

LD-STARS -- long distance sales tool and routing system. There was a series of related projects.

The system allowed to define, store and submit a complicated 800 toll free service.
A call to the same virtual number gets routed to a physical destination based on a number of criteria such as a day of week, time of day, call origination and so on...
UI allows to introduce/modify customer profile with a number of decision trees. Only one tree could be active at a time.
An active tree could be activated -- submitted to an external system, which provides the 800 service.
There was a number of external system adapters implemented (NORTEL/Lucent).
Further system extensions included adoption of VPN, NTS and a voice-routing feature

During my work on the project I have implemented/contributed to most of the backend components, including:

• initial DAO implementation for Oracle
• domain model
• network access layer, based on Berkley sockets
• implementation of gateways to exteranl systems
• ...

An ultra-low-latency market-making application.

The key of the project is a combined HW/SW solution, which goal was to:

• support market-making for a certain set of products
• react to different market events as soon as possible

C++ was an obvious implementation language choice.

The implemented solution made it possible to process market updates:

• keep the current quote;
• produce a new one;
• cancel the current quote, if cannot produce a new one (< 10>

in a matter of microseconds (~4-6).

As a developer on the project I have implemented:

• communication to an external Risk Calculator to pre-calculate price grids for the quoted products
• calculation of a quote, based on the pre-calculated price grid
• safety checker which cancels currently active quote in case of emergency
• and other time-critical modules.

ABBYY FlexiCapture is an Intelligent Document Processing platform built for the needs of today’s complex digital enterprise.

I have implemented an initial version of SDK for FlexiCapture.