Software Development Engineer at Amazon (2022-06 – Present)
Alexa / Voice Commerce and AWS Lambda
- Built and deployed a coordination layer that reduced the time between a voice command and a completed purchase action by approximately 1.5 seconds.
- Designed recovery logic that generates a helpful spoken response when an item is unrecognized rather than terminating the shopping session.
- Developed a dedicated validation service to prevent duplicate customer charges during network retries and system timeouts.
- Integrated a retrieval system that fetches live product details to answer out-of-stock questions naturally.
- Led a training session for 15 engineers to standardize how internal teams integrate voice capabilities.
- Deployed a validation framework that catches voice formatting errors across different hardware devices before reaching customers.
- Configured the automated removal of voice history logs after 30 days to align with EU data retention standards.
- Refactored the initialization process of a serverless runtime to reduce the time a new function takes to start handling requests by roughly 200ms at the 95th percentile.
- Led the migration of internal request routing to a new service that sustained higher traffic volumes with lower operational overhead.
- Mentored 4 junior engineers through operational best practices and internal promotion criteria.
- Re-architected the logging system to redact sensitive customer information from trace files used for debugging.
Senior Software Engineer at Google (2019-02 – 2022-03)
Google Calendar, YouTube, and Google Cloud AI
- Maintained backend services for event management supporting over 500 million monthly active users globally.
- Adjusted database retrieval patterns to improve the speed at which complex recurring meeting schedules are displayed.
- Designed the public interface and permission boundaries that allow third-party scheduling applications to connect safely.
- Managed the synchronization protocol responsible for ensuring calendar changes made offline are eventually reflected correctly on the server.
- Delivered a tool that allows users to export their entire calendar history in a standard, machine-readable format to satisfy data access requests.
- Refactored the service that processes incoming video information updates to reduce the processing time per event by 25%.
- Added a temporary storage layer for frequently accessed video endpoints to reduce strain on the primary database.
- Led the post-incident review following a regional disruption and implemented safeguards to prevent future cascading failures.
- Developed operational runbooks and conducted hands-on training with new hires on managing the service safely.
- Built the automated workflow that permanently removes video metadata and watch history when an account is closed.
- Standardized the process for internal teams to put machine learning models into production to reduce manual configuration steps by 30%.
- Built a connector that allows data analysts to query large datasets using conversational prompts directly within existing analysis tools.
- Investigated and fixed a slowdown in model response time caused by inefficient grouping of incoming requests.
- Engineered the service to automatically failover between geographic locations to maintain availability during zone disruptions.
- Added automatic filtering to the model training pipeline to remove personal identifiers from datasets used in EU regions.
Software Engineer at Cisco (2015-08 – 2019-01)
DevNet and CloudCenter
- Developed the initial platform services that supported the onboarding of over 10,000 external developers in the first 6 months.
- Designed the interfaces that served interactive documentation and testing environments to external partners.
- Implemented a request limiting system during the launch phase to prevent traffic spikes from degrading the platform for active users.
- Added management controls that allowed developers to view, correct, or delete their stored profile information.
- Refactored a reporting module to use cloud provider data sources directly, cutting the time required to generate customer billing reports by 40%.
- Resolved report generation failures for large enterprise accounts by adding intelligent delays and retries to the data fetching process.
- Collaborated with engineers from an acquired company to merge their reporting logic with the existing platform and documented the new standards.
- Documented best practices for interacting with cloud service interfaces to prevent future throttling issues across the engineering team.
- Built a masking layer into the reporting pipeline to anonymize any personal data inadvertently collected before storage.
