Full-stack work in a large monorepo for two core SaaS products: Chikyu (SFA/CRM platform) and ARCH (Notion-like content customization & data visualization platform). Leveraged AI-assisted development tools (Cursor & Claude) to accelerate feature delivery and improve code quality.

• Designed and implemented Zoom Meeting/Webinar integration with secure OAuth and async data sync, eliminating manual data entry and automating attendance tracking for 1,000+ participants.
• Built a high-throughput bulk CSV import pipeline supporting 5,000+ records per batch using S3 and SQS, with real-time progress tracking and idempotent processing.
• Architected an event-driven Process Builder automation tool enabling business users to configure complex workflows without writing code.
• Developed and optimized a fully serverless backend with AWS Lambda, SQS queues, and S3 triggers in a multi-tenant environment.
• Implemented a robust SQS job queue with deduplication, concurrency control, and exponential backoff for high-load reliability.

• Eliminated N+1 query problems in GraphQL resolvers using DataLoader batching, significantly improving query performance.
• Optimized complex database queries with composite indexes to reduce execution time.
• Implemented a reliable background job system with idempotency and retry mechanisms for batch operations including email notifications.
• Designed a direct client-to-GCS file upload solution using signed URLs, removing server-side bottlenecks for concurrent large file uploads.
• Built responsive, maintainable React/TypeScript page flows and reusable UI components from product and design requirements.

Developed core backend services for SOAR (Security Orchestration, Automation, and Response) — a large-scale, event-driven, multi-tenant system processing 50K+ security events daily.

• Designed a task dispatching and coordination system for playbook automation with robust error handling and idempotency, achieving a 99% success rate across 50K+ daily events.
• Implemented the Outbox Pattern to ensure data consistency across PostgreSQL, Kafka, and Elasticsearch, eliminating race conditions and data loss.
• Applied Celery for asynchronous task processing, enabling reliable and efficient execution of security events across the platform.
• Implemented PgPool connection pooling to improve connection performance, reduce DB resource usage, and minimize connection-related errors.
• Integrated an ML prediction pipeline into workflows, reducing MTTR by 20%.
• Refactored the orchestration service using Go's Docker SDK to dynamically spawn isolated Docker containers for secure workflow task execution.
• Built a real-time notification system with Kafka and WebSocket, providing instant alerts and improving analyst response time.
• Applied fault-tolerant patterns (circuit breaker, exponential backoff, idempotent retries) to enhance overall system stability.