• Spearheaded integrations with Python data ecosystem, including Kedro and Pandera, forming basis of composable analytics stack• Led development of IbisML, an open-source library for building scalable ML pipelines that run on any Ibis-supported backend

• Led design and development of Python API for creating projects and defining features in Claypot AI’s streaming feature platform• Contributed first stream-processing (Apache Flink) backend for Ibis, enabling unified stream/batch execution with dataframe API

• Played tech lead for CustomerOne, McKinsey’s end-to-end personalization and customer value management approach and toolkit• Built pipeline deployment component of McKinsey’s machine learning operations platform based on Kedro and Argo Workflows

• Played tech lead for McKinsey’s central COVID-19 epidemiological model, providing clients (US Department of Defense, Commonwealth of Virginia, state of New York, etc.) with data, analyses, and scenarios to help them make informed decisions• Led data engineering teams and mentored junior colleagues and clients on engagements across industries (disaster response and recovery, semiconductor R&D effectiveness, agrochemical fermentation yield optimization, airline crew planning and scheduling)• Led Data Engineering R&D for North America (defined R&D process with other global leads, drafted and reviewed proposals)

• Accelerated unsupervised learning to compute fluid volumes from NMR T1-T2 data using Apache Beam/Spark on Google Cloud• Processed and visualized key information from hundreds of millions of data lake records to help other scientists find relevant data

• Modeled Google Cloud Platform costs as function of data lake user activity to further understanding and inform pricing decisions• Conducted threat modeling assessment of data lake API and Discovery Service; mitigated risk in discovery using Cloud Functions• Replicated CircleCI build environment in VSTS using Docker to support GitHub project in compliance with legal restrictions• Authored Python packages for well and wellbore designation parsing and crawling, parsing, and ingesting LAS files into data lake• Placed third (out of 20 teams) in We Hackathon: Google Cloud Platform Coding Bootcamp

• Developed Elasticsearch plugin for Grafana dashboard and graph editor that supports metrics (not only annotations)• Built process that feeds network operations center portal data into Graphite numeric time-series data store

• Reduced firm’s reputational risk by validating XML files to check data integrity before submission to FINRA• Attained highest level of standards compliance in firm by introducing TDD, Crucible code review, and Jenkins Violations reports• Architected and implemented entitlements framework that supports application- and row-level user entitlements• Designed, developed, and integrated framework that lets users drill from aggregated measures to source system for each metric• Decreased run time of ETL process for AutoSys job run history data from 6 hours per day to few seconds every hour by analyzing scripts called by process, providing faster and more complete, up-to-date data• Developed IBM Cognos report that indicates whether streams, boxes, and jobs met SLA and graphs historical trends, facilitating identification of potential problems; also includes drill-through functionality between levels• Redesigned QlikView summary dashboard to add 30-day trend line and detailed RAG (Red Amber Green) status for each metric

• Built internal website that allows business users to select groups of business entities and their values, thereby removing need for IT to manually analyze subscription requests• Received intensive technical training emphasizing UNIX, C++, relational data modeling, Java, and C#

• Developed website to host feed statistics, service level agreements, and contact information for JPMorgan Collateral Management and Legal Documentation; user interface included datepicker for previous feed statistics and tabs• Wrote program to automatically map JPMorgan GENESIS feeds to JPMorgan GLOBALNET feeds and extract timestamps for these feeds• Created spreadsheets to automatically calculate time difference statistics for feeds, highlight outliers, and calculate new service level agreement times for JPMorgan GENESIS and GLOBALNET

• Wrote program to detect TCP servers on network to identify rogue servers• Demonstrated vulnerabilities via response time covert channels in Xen virtualization platform• Created internal browser-based application for employees to look up network address translation of internal or external address• Used Google Web Toolkit to design linked data explorer that searches online JSON data• Implemented tools to quickly identify suspicious network activity and extract key information from network address translation records• Developed utility to convert data from MATLAB-based format to ARFF for Weka machine learning software

• Identified OCL extended with regular expressions as appropriate language to represent constraints on IDEF1X-based models• Created Infosys InFlux-to-Eclipse Modeling Framework conversion software to support language• Developed part of automatic test data generation suite that interfaces between business rule representation and data generation API