Co Author for a book on Python for Data Engineers and Data Scientists

AI strategy advising, Requirement analysis, AI solution engineering, Proposal writing

I own several open source Big Data and Data Science projects using Hadoop, Spark, Storm, Kafka, NOSQL databases and the related ecosystem. These are some of the important and active projects. They are hosted on github https://github.com/pranab● sifarish Hadoop and Storm based recommendation and personalization engine implementation including content based and collaborative filtering algorithms● visitante Hadoop based web click stream log analytic, log mining and customer and marketing analytic solutions.● avenir Machine learning algorithm implementation on Hadoop, Spark and Storm for exploratory statistical analysis, classification, optimization and reinforcement learning. ● beymani Hadoop and Storm based outlier and fraud detection solution for individual data points and sequences.● hoidla Real time analytic algorithms including outlier detection and approximate queries that can be used by Storm or Spark Streaming● ruscello Time series analytic with Spark Streaming.● chombo ETL solution on Hadoop and Spark, including data profiling, validation and transformation and various utilities

Worked through ThirdEye Data● Responsible for Machine Learning based technical solution for detecting anomaly in power grid system, so that risk can be mitigated before any catastrophic event.● Responsible for design of Tensorflow Deep Learning based object detection system for power line pole and associated structures.● Responsible for design of image bluriness and occlusion detection solutions.

Provide technical support in prospective customer sales calls. Design AI solutions and write technical proposals. Provide technical leadership and direction.● Wrote proposal AI Chatbot solution for enterprise search using Retrieval Augmented Generation based customization.● Wrote proposal for a Deep Learning based predictive model for battery life for Stryker, a medical device manufacturer.●Wrote proposal for Generative AI solution for building design and architecture using text and text to image Generative AI models.● Wrote proposal for a document identification and processing system using Deep Learning for Quest Analytic, a healthcare services company, including feature engineering, model training and deployment ● Designed a Reinforcement Learning based control system and wrote the proposal for a glass coating process for Glas Trosch as part of a solution proposal for them.● Responsible for Tensorflow Deep Learning based solutions for computer vision problems for Southern California Edison, as part of an automated power grid infrastructure maintenance system.● Responsible for building Machine Learning based NLP and search solutions for various Microsoft clients

● Responsible for enhancing virtual event hosting platform with Machine Learning driven match making, personalization and recommendation.

● Responsible for content creation and speaking for webinar series on Machine Learning, including Deep Learning driven Anomaly Detection

AI Driven Real Time Business Operation Monitoring and Risk Mitigation Startup

Work done through ThirdEye Data for various Microsoft customers● Responsible for building Legal Domain specific search engine leveraging Azure Cognitive Search , Azure Blob Storage, Named Entity Recognition, Topic modeling, Semantic Search and Search result re rank learning for European Human Rights Court.● Responsible for a Text Summary extraction solution using Azure Blob Storage, NLTK, ScikitLearn and Gensim for Myriad, Australia.

Responsible for pre sales client meeting, requirement gathering, solution architecture definition and guiding implementation● Enhanced Centerity monitoring platform with anomaly detection capabilities, leveraging my open source Spark based anomaly detection solution.● Architect for a Cyber Security product, leveraging Machine Learning, NLP(NLTK, Gensim), Search Technology, Ontology, JanusGraph, Solr and Python for Merlin International, a government vendor● Provide guidance for a medical image recognition system based on CNN and TensorFlow for Resolve Digital, a startup● Architect for workforce schedule optimizer using Simulated Annealing algorithm for Innova, a government vendor.● Re engineering an IoT and operational data analytic platform for Nimble Storage, a storage manufacturer with Java, Kafka, Spark, Redis and Vertica. Real time ETL with Spark Streaming and analytic with SparkSQL.● Building a multi channel campaign management system with Java, Hadoop and Couchbase for Freedom Pops, a virtual mobile service provider. Optimize campaign with Reinforcement Learning . Machine Learning driven Prescriptive Analytic to improve campaigns.● Reviewed and evaluated Nokia's Big Data and Data Science platform architecture based on Kafka, Storm, Cassandra and Hadoop● Developed Hadoop demo for Microsoft on web click stream analytic to run on Microsoft Azure platform. Used in Hadoop Summit. Developed web click stream analytic map reduce application used in Hadoop training offered by Hortonworks.● Developed course material and taught classes on Hadoop and MapReduce. Taught class in HBase.

● Building retail time series forecasting model to analyze product performance for CPG manufacturers in Python using pandas, fbprophet, statsmodels and scikit

● Formcept is a fast growing start up that offers a Big Data Analytic platform. My role is technical advising in Data Science.

● Designing and building real time data processing pipe line with Kafka as part of integration platform ● Kafka cluster configuration and performance tuning, providing guidance to client application developers. Make Kafka cluster available as a REST service● Designed and implement a log collection system using Syslog, Kafka, Kafka Connect, Splunk and Hadoop.

Architect responsible for migrating existing supply chain software running on traditional platform to a Big Data platform stack, consisting of Kafka, Spark, Couchbase. and Elastic Search● Responsible for defining Big Data platform architecture, product selection and cluster capacity planning and configuration. Designed ETL processing pipeline, ingesting supply chain life cycle events and sensor data.● Designed and implemented a write ahead log based system for handling failure and partially processed messages in the ETL pipeline ● Lead a team of developers to do performance testing of the processing pipeline for throughput on Amazon AWS

● Helping client Nielsen with various Big Data Analytics initiatives● Performance benchmarking of various Hadoop storage formats including Text, Sequence, RC, ORC and Parquet

● Dotstudioz provides a platform to manage, distribute and monetize video content. My role is technical advising in Big Data Platform and Data Science.

● Advising on and building Data Science solutions on Big Data platform for detection of persistent threat.

Member of technical team building ETL and Analytic platform on Amazon cloud using Kafka, Camus, EMR, Spark, S3, SWF, Redshift and Scala for internet TV.● Troubleshooting and code fixing ETL processing workflow running on SWF, EMR and S3● Built an Avro to Relational mapping framework for storing Avro data in Redshift● Built POC for a cloud based log management system with Play, Kafka, Spark and Elastic Search

Lead for building a Big Data platform based on Cassandra, Storm, ActiveMQ and Flume for Equinix, a data center provider. The platform host various Equinix big data applications.● Did capacity planning, configured and deployed Cassandra and Storm clusters. Did design review and provided guidelines for Cassandra data model, Storm topology.● Maintained and enhanced Astynax based Cassandra data services API.

Member of team responsible for Hadoop and Hive based processing platform for financial data. ● Responsible for feature enhancement and maintenance of financial big data processing pipeline built on Hive, Sqoop and Oozie

Key technical member of a team responsible for migrating existing Teradata based Data Warehouse to Hadoop and Hive.● Extended Hive framework with custom RCFile input output format and SerDe to meet update and delete requirements. Essentially added OLTP features to Hive by allowing updates and deletes.● Designed and built a ZooKeeper based Transaction Manager for Hive providing repeatable read isolation.● Tuned Hadoop cluster configuration through exhaustive load and performance testing

Responsible for analytic processing of mobile device usage data using Hadoop, Hive, Hbase and other cloud technologies.● Enhanced Map Reduce jobs for processing large amount of unstructured data from mobile devices and create structured data to be loaded into Hive and other BI systems.● Implemented map reduce jobs and custom input output format in java for ETL processing in large Hadoop clusters. Setup Hive tables for processed data.

These are my spare time activities, primarily consulting for early stage startup.● Consultant, Stealth Startup - Building eCommerce platform for channel management and store front management using HBase and Solr. Implemented search and recommendation functions.● Consultant, Early Stage Startup - Implemented a Naive Baye's data mining algorithm on the Hadoop platform for a customer prospect predictive model.● Consultant, Early Stage Startup - Built an analytical processing platform for product view web log data using Pentaho Mondrian OLAP tools and MySQL. Wrote ETL script in ruby, designed multidimensional OLAP schema and created MDX queries● Consultant, Early Stage Startup - Prototyped customer churn prediction model using Pentaho Weka data mining and machine learning software. Experimented with various classification algorithms and made recommendations.● Consultant, Early Stage Startup - Architect for a vertical search and targeted advertisement platform. Responsible for the core platform on Ruby, Rails and MySQL. ● Consultant, Early Stage Startup - Architect for web site with Spring, Hibernate and MySQL as the platform for a green, transportation related startup.

Senior technical member of Yahoo advertisement platform group.● Back end and web services implementation to support service fees and placement related enhancements

● Built prototype for a large volume billing and financial transaction processing platform for smart grid using Hadoop, Sqoop and Drools rule engine.

Consulting architect and developer for GlobalStar SPOT satellite messenger and tracker related web applications using Ajax, jQuery, JSF, ADF, Java, Ruby, Toplink , JMS and REST.● Worked on next generation SPOT web site architecture using JSF, RichFaces, Hibernate and Spring● Integrated SPOT with a road side assistance and maritime assistance service. ● Enhanced eCommerce functionalities to allow customers to add new services to existing SPOT units and offer new pricing plans.● Implemented various message management functions, including tagging and searching.

Senior member of the architecture team for Oracle Agile's Product Life Cycle Management (PLM) prduct lines. ● Architect for a newly initiated product called Customer Needs Management (CNM). CNM is a social, collaborative system for managing product innovation, ideas and requirement to be built on ADF and Fusion platform. ● Tech lead and architect for an internal web application for development and release process management using Oracle ADF. Using Fusion middleware for legacy integration. Lead a small offshore team.● Built prototypes with Oracle fusuion middleware stack, including BPEL, ESB and BI for eventual introduction into Agile products.● Member of a team of architects, to refactor Agile PLM code, using JDepend, BCEL and other tools for dependency and call graph analysis.

Played a key role as a hands on architect, developer and manager for Manhattan’s J2EE based Distributed Order Management System (DOMS). DOMS is a critical part of Manhattan's supply chain strategy. Responsible for architecting key platform components and managing offshore development team in India.● Designed and built a JSR94 compliant rule engine for various applications including order sourcing. ● Designed and built a business process engine for managing long running business transactions, which forms the core of the order processing platform.● Designed and developed the key components of a global inventory allocation system, which has a highly customizable and configurable feature based on Spring IOC framework. ● Designed and implemented a SOA stack based on SOAP and XML/HTTP. Designed and built an XML document processing framework.

Key contributor to Dorado's PriceMaster product line PriceMaster is a rule driven engine for mortgage loan processing● Implemented critical business features, including sever side java business object and Oracle stored procedures. Made major enhancement to the loan approval business rule engine, including tabular decision table and complex rules● Designed and built a WSDL and BPEL interpreter and a process integration and orchestration engine, which is primarily being used for data loading and state full use case functional testing

Architected and built the J2EE and JMX based management infrastructure for iVAST's MPEG4 streaming media platform● Architected and developed all of the server side for an distributed management platform, based on J2EE, JMX and SOAP, providing management services for a Media Server cluster. ● Designed and built a content distribution and metadata and version management system using FTP, SOAP and J2EE. ● Designed and built a distributed content encoding system which includes a pool of encoding workstations working in a parallel grid model, managed and load balanced by a J2EE server. ● Designed and partially built the server side infrastructure for a Media on Demand service, including Content Access Control, Load Balancing, Edge Caching and Subscription Management

Cedera System is a startup that built a retail supply chain solution. As the Architect and the first technical hire for Cedera, I analyzed requirements, selected platform, technology and architected the key server side components using J2EE technology.● Designed and Implemented an order processing workflow engine. The workflow state transition logic is configurable through an XML file The engine runs in the background communicating through JMS queues. ● Built a rule based order sourcing engine, using optimization algorithm. The rules were based on the proximity of the fulfillment centers, number of shipments etc.

Senior technical member of the ESpeak core group. ESpeak is HP’s strategic Java based EService infrastructure● Played key role in architectural and feature enhancement of the Espeak core engine in the areas of network socket communication and concurrency control. Actively participated in design and code review.

ZipLip.com is a portal for secured Messaging and Collaboration. Left because of company’s questionable future● Provided technical direction in an EJB based product architecture. Built a lightweight EJB Container. Developed a web based secured collaborative application using HTML, JavaScript, Servlet, EJB and JDBC.● Took the lead role in migrating the messaging service to WAP and WML based system for wireless internet access from cell phones.

Lead a team designing and building Onsale’s ECommerce infrastructure with Microsoft technology.● Lead a team designing and building COM/MTS based business components for Onsale’s ECommerce infrastructure.. ● Designed and developed a Java and XML based framework for unit testing COM components . Designed and partially developed a DCOM based server, to manage a multi process Task Scheduler.

Boole and Babbage is a leading vendor for Enterprise management. Left because of a merger.● Designed and developed MTS/DCOM components for managing security and access control for CommandATM”, which is a client server system for managing bank ATM machines.

Responsible for building the core infrastructure components of the “EcoTOOLS” distributed enterprise management product on the Windows NT platform. Left for better opportunity.● Designed and implemented an asynchronous multithreaded Messaging Middleware using MSMQ, COM . Implemented a publish and subscribe feature on top of native MSMQ. ● Designed and implemented a multi threaded Object / Relational Database Server, including object relational mapping, using CORBA, C++ and ODBC.

Lead a s small team, developing a 2 tier Client Server Workflow system for medical application, based on Windows3.1/ODBC/Sybase. ● Designed and implemented software for GUI, database connectivity, report generation and role based screen level security

Responsible for designing and building some of the key components of a MSDOS based PC controlled semiconductor turn key test and measurement instrumentation system in C++, C and Assembly.● Developed GUI, Image Processing and Rendering, Image Database, User Interactive Visual Data and Display software. Designed and implemented a Print Spooler system and Color Printer Drivers.● Designed and developed a proprietary File System for fast real time data acquisition. Designed and implemented a Memory Manager. Developed C++ class library for graphics and imaging.

Responsible for developing mechanical and structural analysis software ● Developed a software package in FORTRAN on Prime 750 / NOS for structural dynamic analysis.