Hi, I'm Pete (Poom)
About
I build full-stack, data-intensive systems that transform complex data into business solutions. My engineering background drives me to break down problems, evaluate trade-offs, and build robust systems—whether in production or at hackathons, where I push my limits and collaborate with other fellows.
I chose software engineering because I love bridging innovation with tangible outcomes. I enjoy the moment when abstract ideas transform into real-world solutions. That’s why I served as a teaching assistant for Algorithms and Data Structures II, simplifying complex technical concepts to help others translate abstract ideas to practical applications.
Outside of coding, you’ll find me playing ping pong or exploring new restaurants across different cities.
Experience
Kinsol needed to transform its proof-of-concept marine diversity classification pipeline into a production-ready system for the first customer demo. I engineered the processing pipeline to ingest raw videos, perform object tracking, and automatically extract highlight frames, enabling a successful live demonstration of the product.
Drove development of the YOLOv8 vision-based pipeline using Python, Parallelization, and Google Cloud securing $30,000 in initial funding.
Optimized annotation runtime by 72% for 30,000+ videos using pre-filtering, distributed computing, parallelism, and GPUs.
Dockerized microservices to stabilize pipeline processing across different machines.
Delivered recurring video analysis using NumPy, Matplotlib, and Pandas to validate AI vision models and improve training outcomes.
Arista required a simplified solution for ISO-based configuration and automated VM first-boot workflows to reduce operational overhead and enhance the customer experience. I built a workflow for ISO configuration validation, a declarative preview system for ISO metadata and contents, and a prototype for declarative ISO generation, improving user interaction efficiency.
Developed workflows for declarative ISO configuration previews and automated ISO generation using Kotlin, Python, and Docker to improve provisioning process.
Enforced Arista network security standards by implementing ISO validation and YANG protocols.
Implemented unit and smoke testing suites with Mockito and Pytest, integrated into Jenkins CI pipelines to ensure reliability of the VM first-boot system.
Visier faced increasing client demand for a scalable data export and a secure solution for managing and anonymizing sensitive customer data. I upgraded the data export system to handle large-scale, multi-file operations and developed a secure de-identification pipeline to ensure data privacy.
Developed a secure data pipeline using Scala, TypeScript, SQL, Docker, and AWS S3 to validate and structure customer data from 15,000+ organizations.
Created CSV/XLSX/Parquet export options with single, multiple, and combined modes to support multi-file download needs.
Reduced network traffic by 90% using efficient zip compression for large-scale data exports.
Built a de-identification pipeline to secure 12M+ records with backup/delete/restore workflows.
VertiGIS needed to ensure that Patch 5.15 was stable and production-ready for the upcoming release of the Geocortex Web/Mobile Designer. I created automated tests for new features and executed comprehensive regression testing to guarantee the patch quality.
Created automated tests with TypeScript, WebdriverIO, and Jest to improve test coverage of Geocortex Web/Mobile Designer software.
Boosted QA bug detection coverage from 50% to 64%, increasing team productivity by 14% during Patch 5.15 regression testing.
Executed functional tests—addressing 30+ issues—to enhance the product quality.
Ruboss’ e-commerce platform (Leanpub) relied on outdated legacy technologies, requiring a next-generation rebuild to strengthen long-term competitiveness. I upgraded the legacy codebase with modern frameworks, making the platform easier to maintain and adapt for future needs.
Migrated Leanpub storefront to modern stacks with TypeScript, React Hooks, EmotionJS, and GraphQL/Apollo to support 2M+ users.
Created reusable React components to reduce redundancy in hierarchical navigation menu.
Integrated CSS and JavaScript code with EmotionJS to increase their readability and reusability.
I was particularly impressed by his structured approach to prototyping—Poom thoughtfully evaluated multiple options and clearly laid out the trade-offs.
Recommendation • Kinsol
Poom... managed the release of a critical feature, followed best coding practices, took on feedback in an agile way, and became an effective software developer.
Final Evaluation • Visier
Poom demonstrated well that he stepped out of his comfort zone by learning new domains, new technology stacks, and different software development practices—all highly focused on quality.
Final Evaluation • Arista Networks
I was particularly impressed by his structured approach to prototyping—Poom thoughtfully evaluated multiple options and clearly laid out the trade-offs.
Recommendation • Kinsol
Poom... managed the release of a critical feature, followed best coding practices, took on feedback in an agile way, and became an effective software developer.
Final Evaluation • Visier
Poom demonstrated well that he stepped out of his comfort zone by learning new domains, new technology stacks, and different software development practices—all highly focused on quality.
Final Evaluation • Arista Networks
Poom performed excellent QA, ensuring high quality for the 5.15 release through meticulous testing... He created several automated tests, many required novel solution designs.
Final Evaluation • VertiGIS
Projects
Awards
