At Media.net, I integrated auction clearing-price signals into an online advertising bidding system, cutting bidding cost by 34.6% while keeping impression volume comparable. I also designed cost-aware bidding objectives and bid calibration methods that improved revenue-to-cost efficiency by 27.3% in online A/B testing.
I built a CatBoost-based auction price prediction model to estimate clearing prices and support bidding decisions when inventory has limited price information. Then I implemented scalable bidding optimization and routing logic, validating gains through offline evaluation and production experimentation.
On the projects side, I built an Anonymous Messaging Platform with a privacy-first, encryption-focused architecture—ephemeral chats, group invites, and support for text, images, voice notes, and documents. I also created an authenticated admin dashboard with message filtering and moderation.
In MedLink Services, I developed a healthcare platform for hospital search, appointment booking, and bed availability tracking, using React and Spring Boot-style RESTful services. I enabled secure access to sensitive medical data shared between doctors and patients.
