XOwned and delivered multiple full-stack features across Pinterest's advertiser data infrastructure. Architected a KV Store data layer serving 1M+ daily requests with a 7x latency improvement (350ms → 49ms p99) and 40% success rate increase. Built event-driven data pipelines using Flink & Kafka for real-time advertiser data propagation. Optimized Airflow DAG execution, cutting pipeline runtime by 50% and reducing cloud compute costs by $12k annually. Operated in on-call rotation supporting distributed databases and microservices on AWS.
Performance gain
7x
350ms → 49ms p99
Efficiency
50%
Airflow DAGs: 5hr → 2.5hr
Infrastructure
60%
Caching strategy optimization
System scale
1M+
KV Store serving advertisers
Reliability
75%
On-call incident frequency
Cost Reduction
$12K
Annual cloud spend eliminated
XLed cost, schedule, and performance of the design, development, testing, and deployment of software and secure cloud infrastructure capabilities for the Space Assets Program Office. Managed $90M in software deliverables and $250M in cloud infrastructure. Rapidly implemented a $50M Agile software re-baselining that averted a 24-month gap in mission capabilities. Ranked #1 of 20 Project Managers in 2021.
Delivery Scope
$90M
Design through deployment
Infrastructure
$250M
Secure infrastructure capabilities
Mission Recovery
$50M
Averted 24-month capability gap
Performance
#1/20
Project Managers, 2021
Impact
24 mo
Timeline Recovered
Access
TS/SCI
Classified Environments
XA full Extract → Transform → Load → Query → Visualize pipeline that fetches baseball player data from multiple sources (ESPN, Fangraphs, Baseball Savant), transforms with normalization, hydrations, and custom valuations (TRP — True Relative Price), loads to PostgreSQL, and serves through Hasura DDN (GraphQL). Visualization layer built with Plotly dashboards for in-season decision making.
XA native macOS application built in Swift/SwiftUI that opens an ESPN Draft Room and scrapes it in real-time for bidding activity and draft results. Concurrently, a separate Python thread hosts a Flask web service where the scraped data is served as a RESTful JSON API — enabling live decision-making during fantasy baseball auction drafts with up-to-the-second bid tracking.
Check out on GitHubArchitecture
2
Swift frontend + Python backend
API
REST
Flask JSON web service
Automation
Real-time
Live ESPN draft room parsing
Frontend
SwiftUI
Native macOS application
