Built and launched a local clothing resale marketplace in NYC from 0 to 1, live on the App Store since August 2024. Learned a ton about React Native, two-sided marketplaces, the cold start problem, and the clothing resale industry.

• Led scaling Dataminr's core Alert datastore on Elasticsearch onto a faster, cheaper, and self-managed hot-warm-cold index to accommodate 5x additional scale.

• Reduced the E2E latency of user’s alert delivery settings changes by 5x by drastically reducing Postgres transaction contention. This decreased the time it took to fully reindex permissions changes for all Dataminr customers from 6h → 1.2h.

• Made significant cost cuts to the teams AWS bill, without impacting performance.

• Fixed complex Java concurrency bugs and connection leaks leading to frequent customer impacts.

• Developed thorough testing framework for the team’s APIs, leading to the discovery of numerous bugs.

• Designed and built ETL pipeline using Hive and Airflow to identify data reclamation opportunities in Salesforce's blob storage cluster (13000 node fleet, 145 PB). With a looming capacity shortage, this tool enabled iterating on analytics data from parsing 16 Tb of log data each day for rapid hypothesis testing. Uncovered a total of 6 months worth of reclaimable storage runway when there was only 3 months of runway left, and reduced additional capacity costs by $4 million+ in 2020.

• Developed scalable Airflow as a Service platform. Built flexible, secure deployment pipelines using docker/kubernetes/spinnaker/terraform for local, stage, and production environments. Successfully onboarded 4 teams to Salesforce Airflow, each requiring it’s own set of integrations ranging from Hive, Presto, Jupyter notebooks, SQS, Spark, and other internal tools.

• Built metrics producer code to send data to Kafka from 30+ teams across Salesforce infrastructure, producing 12 million+ meaningful events per hour.

• Developed real time and batched monitoring applications using NodeJS APIs and React/D3 visualizations for infrastructure teams. Also implemented these monitoring and alerting tools internally for operational management and analysis of our team's rapidly growing ElasticSearch, Cassandra, and Argus (internal TSDB) clusters.

Developed alerting mechanisms and pipelines for the Marketing Cloud's API driven database.

Contributed to the 3 phase development of a monitored, automated, and self recovering database ecosystem. These tools to drastically reduce response and recovery times for incidents at Salesforce datacenters around the world.