Spearheaded development of a new Real User Monitoring (RUM) product stream, enabling customers to gain insights into frontend user experience. This expanded Kloudfuse into a full-stack observability platform and directly supported >$1M in ARR enterprise deals.

Defined product direction: researched frontend observability standards, analyzed competitors, and developed a multi-year roadmap with MVP and milestones in collaboration with CEO/CTO.

Delivered key RUM product features leading a small team of frontend and backend engineers; advanced event exploration and analytics (e.g. segmentation by browser and geography), pixel-perfect session replay, source-mapped stack traces for frontend errors, dashboards, alerting, click-based heatmaps, privacy controls (masking / redaction), and custom data retention policies.

Engineering:

• Multiple iterations of storage backend for RUM events on Apache Pinot; development of advanced custom aggregation functions for time series data (multi-level + filtered aggregations, use of probabilistic datastructures).

• Cloud agnostic object storage backend for Session Replay and Sourcemaps (S3, Azure, GCP).

• High-throughput Go services for RUM event ingestion, enrichment, GraphQL query layer.

• Performance testing framework, including simulator to generate synthetic RUM events at scale, profiling and optimizing ingestion and query paths in both Go and Apache Pinot.

• Integration with platform services including RBAC, ingestion rate controls, and retention management.

Backend: Go, GraphQL, Apache Pinot, Kubernetes, Helm, Redis, Postgres, Python, GCP, AWS, Azure, Grafana

I led critical projects aimed at reducing infrastructure costs, improving the accuracy of search results, and disaster recovery.

• Reduced Big Table storage costs by 10% through purging stale documents, using dataflow, mapreduce, and airflow. Extended the solution to achieve scheduled monthly cleanup eliminating stale documents and correcting discrepancies between search index and source of truth.

• Developed a recovery pipeline to rebuild search Big Table document cache from backups by reindexing historical events at high speed. This was a critical prerequisite for multiple storage side cost saving initiatives planned for the year.

• Migrated search infrastructure components from on-prem to google cloud, removing blocker for Box's exit from on-prem data centers. Migrated use of kafka and zookeeper to cloud hosted equivalents without disruptions to Box customers.

Backend: Scala, Python, GCP (Dataflow, BigQuery, Pubsub), Kubernetes, Wavefront

As an engineer working in the Core Data Demand and Efficiency team within Meta's Core Data organization I worked on optimizing the way product groups consume Meta's online data access and storage systems, safeguarding reliability and availability of these systems from a capacity and security angle.

Optimizations on UDB (UDB is Meta's Global MySQL Fleet):

• Led two major deep-dive investigations UDB that realized a combined 100+ KW (worth USD 1M+) out of a projected 1.8 MW in normalized power savings across the global fleet.

TAO Reliability (TAO is Meta's distributed Graph Data Access Layer fronting UDB):

• Launched a new regression prevention capability that automatically blocked code changes from being committed if landing the changes would significantly increase TAO QPS. Over the course of one year this prevented cumulative regressions totalling > 10% global QPS (4B+), safeguarding TAO from overload scenarios.

• Led a high priority effort to protect TAO managed storage (UDB) from unauthorized mutations made by product owned code in Meta's Hack monorepo codebase by building an extensible enforcement layer around Hack MySQL access APIs.

• Piloted a regression prevention capability to block configuration changes from being committed that potentially regress global TAO QPS.

• Produced and led a 7 hour training course covering TAO, UDB, and the Social Graph with a focus on building expertise on regression triage using Meta tooling, enabling product groups and various Meta teams to up-skill in this complex area. This is crucial to enable a small team of 7 engineers to manage regressions arising from 100s of product groups in Meta.

Backend: Hack (PHP), Ent Framework, ODS, Scuba, Logger, Morphing Framework, Dataswarm, Unidash, various internal tooling and systems at Meta.

As part of the Jobs and Search Platform team responsible for multiple backend system handling job ingestion and search, I helped design and implement several strategic initiatives.

• Designed and implemented features critical to job search integration powered by Indeed: two-way employer and location mapping in search results, enriching Indeed jobs with Glassdoor employer data.

• Reduced delays in job ingestion by 6+ hours by building a system to continuously monitoring job sources

• Reduced expired jobs on the site by 8% by improving expired jobs detection

• Migrating job quality monitoring system from batch to real-time using Apache Storm

• Built a tool to conduct evaluations on expired job detector as part of Glassdoor Hackathon

• Owned production support for multiple job ingestion systems, maintained run books for alert handling

• Performed code reviews

• Worked on multiple micro-services based systems running on AWS utilizing Java 8, Spring, Spring Boot, Project Reactor, Hibernate, Quartz, SQL Server, Redis, Storm, Kafka, Docker

As an engineer on the team owning the Workforce Now HCM product suite I delivered on several strategically valuable initiatives. I was promoted from Lead Application Developer to Principal Application Developer within my 1st year.

• Designed and developed WFN's next generation frontend framework MDF6 based on React. Supported 5+ teams with migrating their frontend features. Mentored engineers on modern frontend development using React, Redux, Sass, Webpack

• Implemented multiple product features owning work across the stack: Quick Hire, Total Source

• Developed API gateway for "WFN On The Go" mobile app to integrate with WFN backend APIs using embedded Jetty server and non-blocking APIs

• Developed multi-tenant persistence framework based on Google Guice

• Established reference implementations for building new REST APIs using JAX-RS and Guice

• Saved 1+ hours of daily developer productivity by improving local application startup time

• Migrated builds from ANT to Maven

• Actively Interviewed full stack engineers as the team was growing

• Presented multiple technical talks at company wide engineering forum

Frontend: React, Redux, Typescript, Javascript, Webpack, Sass

Backend: Java EE (JSF, JPA, JAX-RS, JAX-WS, JAXB, JSPs, Servlets), Docker, Spring, Spring Boot