For official CV/RESUME and experiences, please visit: (https://jupyter.ai & https://verilabs.io). Architecting Realtime Batch & Streaming Data Platforms, Logistics and Pipelines; Architecting their Downstream Analytics Consumption Platforms; Architecting Business Continuity / Disaster Recovery and Governance; etc. On & Premise Technical components include Confluent (Kafka) Platform, Python, Spark, Hadoop, Hive, MongoDB, Cassandra, AWS Glue, Azure Datafactory, AWS Athena, Apache AirFlow, JupyterHub + Docker (for analytics sandboxes), and more.

Director Analytics, Data Standards & Practices reporting to the CDO, Chief Data Office.

INET (formerly known as Island/InetATS at Reuters and now part of NASDAQ), provides the transaction network and engine (a.k.a. ECN) that implements what is universally known as the NASDAQ stock exchange. The core of this platform consists of a grid of 1U systems running a 70mb custom Linux & JRE based O/S; and which work together through a UDP broadcast-based protocol which glues them together. Created instrumentation to measure/identify/improve end-to-end UDP broadcast latencies throughout the ECN platform (including use of network taps, the sock/iperf/netperf/tcpdump/ ethereal utilities, etc.).Developed programs in C to (1) real-time monitor missed packets in the sequenced UDP broadcast stream; (2) to implement a client/server application which provided out-of-band control of servers in the core ECN network.Designed a SAN to capture transactions occurring over the core ECN for historical / trending purposes.Migrated the InetATS DR site from Secaucus, New Jersey to Ashburn, Virginia.

As senior member in the professional services division, provide technical, integration and consulting architectures for business clients that employ SUN Solaris/Sparc-based computer and storage networks as their enterprise-wide client/server platform (including investment banks, telcos, media & entertainment, health care, and other verticals).Details are found here: http://legacy.prismalytics.io

INET (formerly known as Island/InetATS at Reuters and now part of NASDAQ), provides the transaction network and engine (a.k.a. ECN) that implements what is universally known as the NASDAQ stock exchange. The core of this platform consists of a grid of 1U systems running a 70mb custom Linux & JRE based O/S; and which work together through a UDP broadcast-based protocol which glues them together.Created instrumentation to measure/identify/improve end-to-end UDP broadcast latencies throughout the ECN platform (including use of network taps, the sock/iperf/netperf/tcpdump/ ethereal utilities, etc.).Developed programs in C to (1) real-time monitor missed packets in the sequenced UDP broadcast stream; (2) to implement a client/server application which provided out-of-band control of servers in the core ECN network.Designed a SAN to capture transactions occurring over the core ECN for historical / trending purposes.Migrated the InetATS DR site from Secaucus, New Jersey to Ashburn, Virginia.

As a member of a three person Unix Global Engineering team, I designed Sun Solaris-based solutions for Chase’s global business groups. Projects include the design, test and building of a Veritas FirstWatch based HA disaster recovery platform for the Chase private key infrastructure (PKI) using multiple Sun machines connected to D1000s in multi-initiator fashion; SVR4 Packaging of products to be put into Chase's engineering build of Solaris.

Designed and implement all aspects of the compute and storage processing infrastructure (Sun, EMC, Veritas, etc.) that represents the backbone of what serves 4+ million households with the home media service known as the iO.tv (i.e. Interactive Optimum, which is based on a digital set top box that uses http to display a scalable and remotely updatable user interface, and a cable modem to send encrypted digital video on demand, email, and other plug-in interactive services). Designed & implemented the EMC based SAN that services all the functions in the Master Head End datacenter (customer database, digital movie encoders, Conditional access smart card encryption system, set top box auto provisioning system etc). Technologies include ED1032 Connectix; 8730/8430 Symmetrixes; Fibre Zone; Volume Logix; ECC 4.3; SRDF; TimeFinder; Persistent Binding; PowerPath. Designed and built highly available HA clusters based on Veritas VCS and various other availability technologies and best practices (PowerPath/JNI/Qlogic, DMP, IP multipathing, mirroring, load balancing through Resonate etc); Designed and coded a black box engine that sits on top of Sun's jumpstart product that eliminates all of the traditional shortcomings of jumpstart and makes it enterprise scalable. I evaluated and installed the new SunFire 6800, 4800, 3800 products; the evaluation included creating single and dual partitions and the domains within them using the System Controller Console and Domain Console CLI; physically adding and removing components like CPU/Memory, Interface Boards, PCI cards in one domain to verify the isolation in other domains; flash upgraded the 6800 & 3800 firmware; extensive load testing including designing and creating a platform for Oracle 8.1.7/9i Parallel Server testing on EMC storage. Essentially, designed and implemented the back end (Head End) computing infrastructure that serves millions of customers in the NYC-TriState area.

Technical-level role involves the design and implementation of Big-Data infrastructure components to institute a near real-time publications Recommender Engine for clients/subscribers of it's financial research publications line-of-business. Technologies include Cloudera Hadoop, Mahout to generate item-based, item-similarity, and item-cluster driven recommendations, and Apache PIG & Python for Map/Reduce jobs including pre and post processing.Executive-level interfacing role involves managing customer expectation and keeping the project on track.As part of an evaluation process to identify successor technologies to Bank of America's current real-time stream processing system, implemented a P.O.C. based on (Twitter's) Storm to process over a million FIX messages per second: Built the 1-Nimbus + 3 worker-node Storm cluster (including fail-fast provisions); designed and wrote the Topology and Spout & Bolt components (inJava) that solved various high speed stream processing use cases, including subscribing to their trade message stream topics, performing operations on received messages as Storm tuples, re-publishing the results back to the stream as a (possibly different) topic, and on a limited basis wrote to a Cassandra database as well.

Retained in a consultant capacity by NYMEX to fully design and implement a multi-site SAN (Storage Area Network). Other than my own, no vendor or third party professional services were used in any of the work to be described. Please refer to the following technical diagrams and papers, which relate to the work described below:http://www.prismalytics.com/cv/doc4.pdfhttp://www.prismalytics.com/cv/Buffer_Credit_Calculations.pdfSuccessfully designed/specked out/purchased/installed/configured the following: (1) Took two unused Hitachi HDS9960s, moved one to a DR site, and got TrueCopy working between the two via direct FC-AL over two DWDM lines. This step included upgrading HDS microcode, provisioning disks via the SVP and Remote Console, creating an RCU, ensuring different DWDM paths etc. (2) Based on the distance between the two sites, calculated the BB & EE CREDIT requirements and procured and configured 4 McData ES3232 switches (a redundant pair for each site); through a pair of local ISLs between the two switches at the same site, and pairs of DWDM ISLs between the two sites, converted FC traffic from just TrueCopy traffic over DWDM to general FibreChannel SAN traffic over DWDM (which included TrueCopy traffic) (3) Purchased Hitachi SANtinel licenses for Lun Masking (4) as the first clients of the SAN, created a mission critical VCS 3.5MP2 cluster (for a mission Critical Oracle Database) on at one site, and synchronously TrueCopy replicated the data to the DR site. Other accomplishments included creating specialized High Availability storage solutions using multi-initiator configurations using the D2 StorEdge array; configured Squid Proxy servers; and other Sun/Solaris architectures/implementations.

Extended NetApp’s basic enterprise monitoring platform (known as Operations Manager (OM) / Distributed Fabric Manager (DFM)), by writing code to (1) tap into notification events generated by this products and, where necessary, invoke lights out corrective action (such as just-in-time growing of volumes that reached capacity) and (2) use DFM as a gateway/portal into the NetApp ONTAP enterprise, to generate custom dashboards (in HTML or CSV format) to report the health of an entire NetApp enterprise. Also perform heterogeneous migrations for their customers using tools like SecureCopy, Beyond Compare, Rsync, etc. to NetApp filers (Storage Controllers) from competitor storage platforms.

Storage, SAN, and Sun/Solaris Engineer designing and implementing the infrastructure for New York City’s employee time entry and management portal. The project objective was to implement a web-based application that is used by employees of New York City’s 129+ agencies to enter in hours worked, vacation time, etc. The new infrastructure replaces numerous older mechanisms for entering and tracking such data, with a central modern system, making the process more cost effective and forecast budgeting easier. Technologies include, SunCluster 3.2; VCS 4.x; EMC Symmetrix; Clariion CX300; IBM application server products (MQueue; WebSeal; WebSphere; TIM/TAM), etc.

Engineered, built and administered the EMC-based Storage Area Network on which Mail.COM hosted its critical businesses: consumer, partner ISP, and business-to-business messaging (for example: Mail.com, Email.com, Iwon.com, CNN.com etc.). The design consisted of four EMC Symmetrix 3930s, two 3830s, three EMC dual director Connectrix Fibre Channel Switches, four EMC Celerra File Servers, and a dozen 28CPU Sun UEx500 Enterprise Servers. Additional components include software components such as EMC PowerPath, ECFM Connectrix Manager, Volume Logix, ECC SymCLI & OSM Manager, Veritas Volume Manager (VxVM), Veritas FileSystem (VxFS), etc.On the SAN side of the design, considerations such as redundancy and resiliency were made possible through multiple active (Power) Paths across different Fibre Channel switches, JNI HBA boards across different Sun I/O boards, VxFS for quick filesystem recovery, and VxVM mirrored O/S and SWAP volumes. High availability considerations were accomplished through the configuration of an additional UE6500 with visibility to the entire SAN storage pool via Fibre Channel zoning and Volume Logix HBA to FA hypervolume authentication. In this configuration, the UE6500 can assume the compute task of any of the other N UExx00 class Sun servers through sd.conf management, and through the VxVM diskgroup import/deport mechanism. ECC SymCLI TimeFinder and BCV’s are also configured for 3rd mirror based data backup, snapshots etc.Based on Sun’s Jumpstart model, engineered a platform for jumpstarting machines on Mail.COM’s global networks, without the need to configure jumpstart or even boot servers on remote segments. Traditional jumpstart environments minimally require a boot server on each client network segment. A result of this Light Weight Jumpstart Server design is that network segments no longer require (and are prohibited from having) their own boot, configuration, and install servers.Complete detail here: http://www.ComputingArchitects.Com

Engaged to conceived, product design and software develop an end-to-end, custom information processing factory to enable Lab Analysts to identify, validate and forecast global business and economic trends of consequence. Python-3 and related frameworks are used to build the factory, while various enterprise distributed databases are used to store output Datasets (via SQL and ORM means) in a browsable Data Catalog. Processing features include: applying Taxonomy tiers to the data; Matching data attributes to Master Reference Data; Routing Data Quality exceptions to remote teams for reconciliation; End-to-end Traceability via step-by-step logging and correlation; and more. This build-to-fit custom factory accelerates Lab analysis projects by alleviating Lab Analysts from having to perform painful, rote, error-prone data processing tasks, and enables them to instead focus on the financial interpretations of it.