Engineered event-driven data pipeline, synchronizing millions of records a day from different data sources to GoFundMe's search engine through a distributed worker architecture (MySQL, SQS, Celery, Django).
•
Reworked the textual and geo search result ranking configuration, boosting user click-through and conversion rates across GoFundMe's search result pages by 12.9%
•
Developed various RESTFUL and GraphQL APIs in Django, Laravel, and Kotlin Springboot across GoFundMe's homepage, search page, and campaign page surfaces.
•
Migrated the continuous integration and deployment pipeline for several python microservices from Jenkins to Github Actions, resulting in markedly faster and more comprehensible builds and deployments.
Developed ETL functionality to automate a complex manual process of data collection and cleaning (Java, C#, SQL)
•
Standardized customer data relationships and granularity within a centralized data warehouse (SQL Server)
•
Developed Application Database and RESTful API to manage users and report groups within customer facing web portal (C#)
•
Authored and embedded a series of business intelligence reports within customer portal; implemented dynamic, multi-role, row level security to isolate customer data (Power BI)
Developed Printer Diagnostic Automation tool which links technicians to relevant trouble- shooting documents based on the error/jam code reported by a malfunctioning printer (Java, SOLR)
•
Developed autonomous Crawler and Entity Extractor to keep the Printer Diagnostic Automation tool up to date with newly released articles (Java, Python, SOLR)
Used context-based keyword analysis to classify expected routing of regulatory mail for a financial firm – done to automate and digitalize a manual, paperbased regulatory process (Java, SOLR)
Legal Document Search Tool:
•
Prototyped search tool for legal firm optimized to locate specific court cases based on judges, plaintiffs, and defendants involved in each case (Java, SOLR)
Canon Search Engine Evaluation:
•
Evaluated Canon’s Search Engine and used evaluation results to improve accuracy from 51.55% to 82.71% on a 13,000 query testing set. (Python)