Arturas H Vaitaitis
(Current status: passive interest)

linkedin/varturas

North Bergen NJ,

Contact me


Summary
Senior data scientist and quantitative technologist with experience in building and managing front office, trading and market data systems and groups, expertise in algorithmic and high frequency trading, statistical data analysis and system programming; Ph.D. in particle physics; Extensive leadership and problem solving skills.

Technical skills
Management Data ScienceSoftware
Building trading and market data group from grounds up, heading front office development, overseeing IT enterprise architecture, managing performance improvements in electronic trading systems; project and development management
  • Java ( Core 6 yrs )
  • C++ ( 7 yrs )
  • perl (9 yrs experience)
  • C (10 yrs )
  • FORTRAN ( 8 yrs )
  • Linux, Solaris, Windows, embedded devices
  • TCP/IP, RPC, tcpdump, traceroute...
  • Posix threads, unix IPC
    • R ( 5 yrs, hypothesis testing/regression,purrr,tidyverse,knitr )
    • kdb+/q ( 7 yrs )
    • python2.7 ( 15 yrs, pandas,numpy,stats,ARFIMA,GARCH )
    • python3 ( 3 yrs )
    • matlab/SAS ( 1yr)
    • Deep knowledge of Fixed income and equities products
    • Understanding of stochastic calculus,
      statistics and finance. Ph.D.

    Experience Summary
    Cantor and BGC companies (BGC,CCA,Aqua)                                                                 05/2010--present
    • Managing director and CTO of 2 Cantor startup companies
    • Head of data solutions and senior quant on a business side
    • Build from ground up market analytics group
    • Architected book building framework in kdb
    • Served as a problem solving ¿swat¿ team across Cantor subsidiaries
    • Developed novel TCA/impact model for UST
    Experience
    Cantor Fitzgerald, New York City                                                                                       01/2018--present
      Senior quant, head of data and analytics solutions
      Technical Environment
      kdb+/q, R, Python
    • Develop volume profile and volume prediction python application for live trading
    • Research and tuning of internal high-touch algorithm parameters
    • Develop post trade (TCA) reporting framework in R-markdown - R, kdb+/q, R-markdown and latex
    • Building tick-data plants for a algo trading system for orders, executions, ref data, market data and analytics
    BGC Partners, New York City                                                                                           08/2015--01/2018
      Head of market analytics group
      Technical Environment
      kdb+/q, R, Python, Java
    • Designed and develop centralized tick-data storage for the whole company and subsidiaries
    • Designed architecture for Data Complex, spanning event tape storage (HDFS) and stream processing
      (Apache Spark, Kafka), operational health and DevOps dashboard. Developed event tape and stream processing components
    • kdb+ ticker plant, RDB/HDB framework hands on development
    • Develop bond analytics, loss ratio, fill ratio reports for broker desks
    • Designed permission access control system based on Delta Control product
    • Designed and developed trading system benchmarking and reporting based on kdb+
    • Designed and develop benchmarking and performance measuring tool for internal trading systems
    • Developed research framework based on kdb+/R/python, visualization in shiny
    • Integrated with ELK (Elastic-Logstash-Kibana) as visualization dashboard/search tool
    • Day to day management of data group, administration, documentation, hiring and operational support
    Aqua Equities, Cantor / BGC startup, New York City                                                   08/2013--08/2015
      Managing Director, CTO of dark pool, block equity startup
      Technical Environment
      Linux,Mysql,Java,.Net/C#,python
    • Introduced processes and procedures for setting priorities, managing projects, best practices and support (Jira, agile process)
    • Conducted trading data collection and trade performance analysis (java, python)
    • Designed the entire trading system from un-scalable to horizontally scalable in trading volume and number of clients. Developed high level architectural design and low level component design. Oversaw implementation and execution. (java, LMAX disruptor).
    • Designed and developed UI desktop installation and watchdog framework that controls trading UI remotely (C# and java)
    • Develop and support connectivity to various client OMS systems (McGregor, Newport, etc.), (java)
    • Designed, develop and test an interface and connectivity to Triton EMS (ITG) (C# and java)
    • Day to day management of development group, administration, hiring and operations support
    CCA investment advisor, Cantor Fitzgerald, New York City                                         08/2012--07/2013
      Managing Director, Cantor investment advisor, startup CTO
      Head of technology at Cantor equity trading startup
      Technical Environment
      Linux,Sybase,Java(Hybernate),C++,python
    • Design and development of market engine, a critical component to hedge market exposure in real time and to manage risk (python and Java)
    • Develop order routing and execution services interface (c++/c), migrated to Quickfix/Java implementation
    • Conducted trading performance analysis and strategy research, modeled adverse selection premiums (excel, python and R)
    • Day to day management of IT group, project management, technology purchases, vendor and client communications, hiring, administration and trade support duties
    BGC Partners (Cantor Fitzgerald), New York City                                                       03/2010--08/2013
      Global enterprise architect
      Mission: Overview of firm-wide technology architectures, provision of guidance and recommendations, implementation and execution of key recommendations, "fire-fighting" the problem areas
      Technical Environment
      Excel (Bloomberg DDE), .Net, C#, VBA, IronPython, C++, eSpeed API
      Report directly to CIO
    • Oversight of Static Data for BGC middle and back offices. Worked in a capacity of a data governance head.
    • Oversight of front office technologies and operations for BGC brokerage; Worked in a capacity of product and project manager and, for 2 months as a development manager; Was involved in project prioritization and day to day management of the front office development group; My recommendations included personnel relocation closer to desks, technology and network architecture changes; As a result, number of serious production incidents become negligible
    • Conducted data analysis to verify curve tracking algorithms deployed on trading desks. Drafted a proposal for a rapid application development environment and oversaw it's implementation
    • Managing a program of projects to modify BGC IT to comply with Dodd-Frank regulatory legislation (aka Wall Street reform).
    Sun Trading, Chicago                                                                                                         04/2009--03/2010
      Head of Market Data
      Technical Environment
      postgresql, kdb+/q, C++, python, perl, R
    • Establishment of a market data group, from the ground up. Managing all data aspects, including architecture and data base design, research and development, hiring and vendor relationships
    • Supervised establishing exchange connectivity to a wide variety of exchanges; Created a prototype for FAST parser, designed architecture for capture and processing of exchange feed data and loading into a time series database kdb+
    • Design and creation of time series database (kdb+) to store market tick data, schema design and optimization
    • Establishment of tick capture with precise timestamps, using low level Unix utilities, Kernel hacks
    • Creation of data quality management system, data cleaning procedures, user notification framework
    • Measure and report data integrity and latency
    • Monitor time precision of feed capture processes by measuring realtime clock offsets of colocated servers, using distributed framework of time servers to multicast and record timestamps and roundtrip time. Clock offset calculation have been done offline during data analysis
    • Design, architecture and implementation of Security Master database that keeps track of corporate actions, symbol changes, futures rollover and contract expirations.
    • Setup data access layer to research frameworks, including R, python and C/C++ applications used by researchers, financial engineers and traders
    • Design and develop kdb scripts library and repository, implement build and release procedures
    • Establish a prototype for near real time data processing and loading used in live trading
    Morgan Stanley, New York City                                                                                              09/2006--04/2009
      Vice President, Equity Trading Lab (ETL), Market Data Group
      Technical Environment
      Sybase, perl, SAS, R, C++, kdb+/q, python
    • Researched, designed, developed and turned-over to production News Feed processing modular system (written in perl and SAS) to parse, extract and analyze earning announcements and company issued guidance from Dow Jones News, Internet and other sources for Market News. The processing infrastructure consists of three logically separate perl processes (parsing, extracting referential information and merging different sources), with intermittent data transparently mapped on directory tree
    • Research arbitrage value of data contained in financial news: blogs, message boards and internet news feeds. Developed back-test system (written in python) to convert sentiments into simulated trading list and analyze overall performance and P&L. Conducted statistical analysis (time distribution, regression, correlation with returns etc.) using R and SAS
    • Designed and developed data correction system capable of repopulating market data in production databases from different pricing sources. System does 4-way comparison of data sources to detect errors and discrepancies, capable of rollback and insertion of new records. Correction is applied to production database (Sybase) and to flat files. It is written in perl
    • Design and implement alerting system for market data quality monitoring, crucial to development of algorithmic trading strategies. Alerts dispatched by a perl-based daemon (monitored further by autosys) via email and Netcool and are linked to appropriate web-based (wiki) documentation. System monitors over 1000 production jobs, highly customizable and reliable
    • Conducted firm-wide third-level on-call support of market data
    Bloomberg Financial Markets LP, New York City
      Technical Lead/Senior Software Developer — (3 direct reports)                                        03/2000--07/2006
      Technical Environment
      C++, perl, MySQL, Java, C
    • Research the concepts and Designed unique system to monitor latency and integrity of market data. Created the prototype that worked on a generic market data format. The prototype method finds a time-series match between the data stream from the source of market data and from the client, by maximizing the cross-correlation value concurrently calculating integrity and latency (Patent application pending).
    • Conducted Vendor/Product research of Search Technologies as a part of participation in a working panel of company wide experts.
    • Assumed the project manager role to lead implementation efforts by other groups, integration, testing and rollout of beta version of the market data monitor based on the prototype design described above. Productized the monitor via integrated Enterprise Management System (EMS) framework on the basis of the third party product, HP Openview. EMS is a framework to manage Bloomberg applications.
    • Research concept of generic correlation of disparate EMS alarms coming from different applications to automate problem finding system. Implemented mechanism to route all EMS events to HP Openview system. The mechanism provides enhanced event categorization as well as filtering and rate suppression capabilities to the console room Operations. The legacy system used to rely on many operators noticing a reoccurring alarm in a scrolling stream of alarms in a window. The new system supplies a ‘manage-by-exception’ capability to an operator thus drastically improving problem detection ability.
    • Created an application for automatic scheduling and handling of batch system commands, for network troubleshooting (backend written in C++, front end HTML). The library is used by other developers in R&D to troubleshoot network problems replacing the direct access granted to Tech. Support, thus closing a potential security hole.
    • Designed and developed multithreaded (consumer/producer) real-time high-performance (500 Kbytes/sec) CORBA based middleware (written in C++). It served as a backend metrics collection system for automated outage detection system described below.
    • Conducted research, collected statistics and analyzed the patterns of network outages. Researched, designed and developed the prototype for correlation analysis of network outages. The automated outage detection system determines the most likely point of failure in the network before customers could take a notice. This analysis provided vast capabilities for real-time troubleshooting network failures. Researched and developed the model of background noise levels present in measurements of network outages. Used the model to implement noise from signal subtraction. Transferred the prototype design to other group members for the implementation. Participate in implementation process.
    • Created a tool for handling and scheduling automated batch commands (ping, trace route)
    • Designed and developed an allocation algorithm for load balancing of Bloomberg clients amongst backend machines and gateway servers. Implemented (perl cgi) and turned over to Operations an application based on the algorithm. It is widely used by operations and monitored as a critical.
    • Developed complete (backend C++ and front end Java/JavaScript) application for troubleshooting and monitoring voice-over-IP communications over Intranet
    • Development of a data collection system (C++) that probes Bloomberg Internet and Intranet for network metrics (IP, retransmition rates, network delays).
    • Optimized and normalized the proprietary database based on UNIX indexed hash tables (C++) used for storing metrics for near-real time troubleshooting. Optimized the database to be able to insert up to 10,000 records per second. The resulting system was unique to the R&D department (perhaps in the industry) in efficiency and speed
    • Implementation of a front end Java client to display analysis of network performance metrics in a form of graphs and/or tables, grable (Java, Javascript and fastcgi). It had been also used for logging the usage of financial functions (stocks, bonds and indexes) on Bloomberg application.
    • Expanded the data collection system from being solely proprietary database to the Open Source MySQL database on Linux. This greatly increased our flexibility for tuning and optimization of the data collection to increase its performance up to 3,000 inserts per second. The system probes Bloomberg Internet and Intranet for network metrics.
    • Developed and maintained Unix C++/STL development libraries for network name lookups, IP resolution, address translation and API for CORBA (omniORB 3.0) widely used by back end server applications
    • Created an cgi perl-based point and click user interface for the 'intruder alert' web based tool. It had been used by the security team as a trigger mechanism to start probing our network for security breaches.
    • Performed production LINUX (redhat7+, 2.4+ kernel) system tuning and upgrades
    Nevis Laboratories, Columbia University, NY Reasearch assistant                                                         10/98--03/2000
    • Upgraded and implemented a complete statistical Monte Carlo simulation of particle interactions in the detector hardware, their conversion to electrical signals, simulation of electrical and electronic noise. Used Monte Carlo simulation to compute and subtract noise from the measured signal
    • Analyzed and assessed sources and levels of systematic and statistical errors applicable to measured results
    • Conducted a novel research in particle physics, results are published in scientific journals
    • Supervised the research work of 3 students
    Fermi National Laboratory, Aurora IL Reasearch assistant                                                                           04/95--09/98
    • Designed and developed software (using Fortran77) for the reconstruction and analysis of experimental data recorded onto tapes during experimental run
    • Ported analysis software from SGI platform to Linux, created new set of configuration and makefiles, researched compiler behaiviour. Conducted futher data analysis on Linux cluster
    • Designed and develop software for the reconstraction and analysis of the experimental data
    • Structured and organized storage of data recording during the experiment run
    • Customized graphics program( event display) that allowed real time visualmonitoring of interaction of particles in the detector
    • Performed calibration and tuning of the experimental setup
    • Arranged the assembly and testing of the data acquisition hardware
    Columbia University, New York, NY Teaching assistant                                                                             09/94--03/95
    • Instructed and supervised students at the physics lab;
    • Developed computer simulation of particle interactions for an existing hardware detector setup

    Education
    Columbia University, New York
    Ph.D. in Experimental physics                                                             03/2000
    Master of Philosophy                                                                           10/99
    Master of Science                                                                                11/96

    Relevant Coursework
    • Software Engineering (in Java)
    • Data Structures and Algorithms
    • C++ programing language
    • Object Oriented Programing and Design
    Moscow Institute of Physics and Technology, Moscow, Russia
    Master of Science with Honors                                                             8/93

    Honors/Awards
    • 2nd place in national Lithuanian physics olympiad
    • Columbia faculty fellowship

    Areas of Interest
    • Automated news comprehension
    • Algorithmic trading and behavioral finance
    • Physics of Time
    • Knowledge and Information Management
    • Writing
    • Traveling
    • IoT
    Talks
    • Study of the inclusive production of neutral pions in pp interaction, using the reconstruction of the pi0--decay. Workshop on Physics Programm at UNK Serpukhov accelerator, Protvino, Moscow region, Russia
    • Searches of a heavy neutrino, New perspectives 97, Fermilab, Illinois
    • A Searche for Heavy Neutral Leptons in the NuTeV (E815) Experiment at Fermilab 1998 Joint APS/AAPT Meeting, Columbus, Ohio
    Publications
    • SEARCH FOR NEUTRAL HEAVY LEPTONS IN A HIGH-ENERGY NEUTRINO BEAM A Vaitaitis, RB Drucker, J Formaggio, S Koutsoliotas¿ - Physical Review Letters, 1999 - APS
    • TESTS OF A CALORIMETRIC TECHNIQUE FOR MEASURING THE ENERGY OF COSMIC RAY MUONS IN THE TEV ENERGY RANGE. NuTeV/CCFR Collaboration (A.P. Chikkatur et al.). Z.Phys.C74:279-290,1997
    • HELICITY EFFECTS IN NEUTRAL HEAVY LEPTON DECAYS. By Joseph A. Formaggio, Janet Conrad, Michael Shaevitz, Artur Vaitaitis, Robert Drucker (Columbia U.). 1998. Published in Phys.Rev.D57:7037-7040,1998
    • MEASUREMENT OF SIN**2 THETA-W FROM NEUTRINO NUCLEON SCATTERING AT NUTEV. By NuTeV Collaboration (K.S. McFarland et al.). Mar 1998. 12pp. Talk given at 33rd Rencontres de Moriond: Electroweak Interactions and Unified Theories, Les Arcs, France, 14-21 Mar 1998. e-Print Archive: hep-ex/9806013
    Patents
    • SYSTEM AND METHOD OF DETERMINING DATA LATENCY OVER A COMMUNICATIONS NETWORK — <US Patent[8,681,643, 2014],(11/696,864)>
      The proposed method finds a time-series match between the data stream from the source of market data and from the client, by maximizing the cross-correlation value concurrently calculating integrity and latency.
    • Method and system of attachment and detection of attachment of a wearable sensor to clothing material AH Vaitaitis - US Patent App. 14/578,904, 2014

    Language ability

    • Fluent in English, Russian, Lithuanian