Contractor2017 - PRESENTIndependent Contractor
Technologies: MongoDB, Neo4j, Node.js, Amazon Web Services (AWS), Apache Kafka, Python, Solr, Elasticsearch
- Designed a new search engine tailored for a identity management company to implement new business logic and find abnormalities in the data and data usage.
- Created new graph model design for the identity management company to create targetable and tailored experiences for their users and also help with fraud detection.
- Wrote an ETL process with Python to get the data from the MongoDB to feed into Elasticsearch and Neo4j in real time.
- Developed an alert system on Elasticsearch and Neo4j for data abnormalities and graphs so they can visualize the data; it sends a daily report to the executive team.
- Built a Developer API portal to expose all the functionalities to the developers.
Architect, Platform Data Services (PDS)2015 - 2017Yellowpages | YP Holdings formerly AT&T Interactive
Technologies: Mesos, Docker, Java, Elasticsearch, Python, Apache Kafka, Cassandra
- Designed a new Kafka/Spark infrastructure for the platform and data services team to migrate from a legacy UDP protocol (LWES, lightweight event system).
- Implemented monitoring for the in-house cloud application that was running on Mesos; used Systemd, Kafka, KairosDB with Cassandra as a back-end, and Elasticsearch.
- Built custom metrics and monitoring platform around the new Kafka and Spark infrastructure.
- Extended the above solution for a company-wide monitoring and reporting solution for all the apps deployed on the internal Mesos cloud infrastructure using Systemd, Collectd, Logstash, Kafka, Elasticsearch, and Grafana.
- Developed prototypes to get rid of New Relic and Splunk; the aim was to reduce cost and unify the monitoring platform for YP.com.
Architect, Consumer Platform2012 - 2015YP.com | Yellowpages.com
Technologies: Java, Solr, Elasticsearch, Memcached, MySQL, Oracle, Cassandra, Ruby
- Migrated an antiquated Perl-based system to Solr/Cassandra (a high performing scalable system) for the GIS project so that we didn't need to rely on Bing and Google maps for geocoding and reverse geocoding needs and save the company million dollars in expense.
- Developed prototype replacement ETL pipelines for the data team. The goal was to match and merge data from different sources based on a multitude of business rules and cut down the processing time to a couple of hours instead of 24 hours.
- Designed and implemented a tracking system from user queries origination all the way back when the results are displayed. This gave us a way to profile each system in the stack for a performance analysis of each system in the stack and improve on them, one layer at a time.
- Led the API team to reduce the latency middle layer Ruby application by 50%; we looked at the data that we could gather from the above systems and redesigned the stack and underlying systems to increase the throughput and decrease latency.
- Built a consumer OPS team that developed monitoring and alert strategies.
Architect, Search Platform2010 - 2012YP.com | Yellowpages.com
Technologies: Hadoop, Solr, MySQL, Java, Elasticsearch, Cassandra
- Designed and built new systems to modernize and increase the usability of Yellowpages' properties.
- Developed a personalization engine that tracks millions of user behaviors on Yellowpages properties. It suggests results, based on the user’s activity, and activity to similar users in a similar geography.
- Built a geo-aware autosuggest system; the system returns results based on geography and recommends local businesses that are open at the time of the search.
- Created an autosuggest system for Yellowpages' properties.
- Developed a custom spellchecker based on YP data and user behavior.
Search Operations Admin2007 - 2010YP.com
Technologies: Bash, Python, Java, Solr
- Scaled the existing search systems, automated the deployment process, and increased uptime to 99.9999%.
- Designed and developed City State (geolocation lookup, validation system) in Solr.
- Designed data ingestion pipelines and search modules for categorization (classification of user input queries).
- Created the Solr infrastructure design (LEARCH, local and search combined) the primary search engine for Yellowpages.
- Migrated the search engine from Fast to Solr which would save millions of dollars of licensing fees.
- Modified and developed document and query pipelines; Fast-based.