Architected and developed a scalable e-commerce analytics platform using Django, Celery, and PostgreSQL, focusing on high-throughput and low-latency data processing to meet the demands of a growing user base. It creates more than 1 mln database entries per day and serves 1000+ users daily.

더보기

• Led frontend architecture and development for a digital asset trading platform handling approximately $50M in monthly transaction volume. • Designed and deployed a low-latency real-time data layer using custom subgraph indexers, graph-node, and a dedicated REST and graphQL API, ensuring latency under 200ms for trading dashboards and partner integrations. • Technologies: Next.js, Node.js, TypeScript, Web3, AWS, GCP, Docker, GraphQL

더보기

Contributed to B2R2 binary analysis framework. Implemented instruction lifters and format parsers for MIPS, RISC-V, and PA-RISC, enabling support for additional CPU architectures in reverse engineering tasks.

더보기

• Co-architected, built, and scaled the core backend analytics platform, processing over 5M daily transactions and supporting 30+ retail clients serving a combined 3M+ daily active users. • Reduced report generation latency by 60%. Achieved this through database schema redesign, table partitioning, advanced SQL optimization, Redis caching, and scheduled jobs. • Built real-time dashboards and launched an AI-powered Collaborative Filtering recommendation engine generating 100k+ user interactions - boosting client conversion rates within weeks. • Technologies: Django, Next.js, TypeScript, PostgreSQL, AWS, Redis, Pytorch, Docker

더보기

• Developed and trained a conditional GAN model for grayscale image colorization, increasing the usable training dataset size by 30% for downstream computer vision tasks. • Architected and deployed a scalable web crawler, collecting over 1 million images for training AI models. Achieved 100,000+ images/day throughput and 95% success rate. • Engineered a high-performance web-based 2D data labeling tool, accelerating the creation of large datasets for AI models by 10x. Implemented intuitive features like keyboard shortcuts and temporal navigation, reducing average labeling time per item by 80%. • Technologies: Python, PyTorch, AWS, MongoDB, Django, JavaScript, Apache Kafka

더보기

Refined real-time delivery tracking system, enabling dynamic courier location updates for customers, resulting in a ∼30% reduction in customer service inquiries and fostering a 15% increase in repeat order rates. •Implemented a dynamic ETA calculation engine, which utilizes real-time traffic data and driver behavior models to provide customers with more accurate delivery times, reducing average customer queries by 30%.

더보기