Konstantin Kriviakin

As a Technical Leader at Meta, I drive the development and optimization of ad targeting systems, connecting over 3 billion daily users with relevant advertisements. My role involves solving complex technical challenges and leading a team to innovate in ad delivery, leveraging AI and machine learning to enhance audience selection and campaign efficiency. I have also been instrumental in improving developer productivity by creating tools and frameworks that streamline work with our extensive C++ codebase, benefiting thousands of developers across the organization.

Key Achievements:

• Advantage+ Audience Product: Played a pivotal role in the development and success of the Advantage+ audience product, which now accounts for 25-30% of Meta's revenue, by innovating in ML model evaluation and ad filtering infrastructure.
• Efficiency Improvements: Established the Ads Delivery Efficiency area, achieving a more than $100 million reduction in Capital Expenditures over three years by optimizing server capacity and enhancing algorithm efficiency.
• Developer Productivity: Led the Targeting platformization project, saving over many Engineer-Years annually by streamlining the codebase and developing tools that enhance productivity for thousands of developers.

As a Principal Software Engineer at Tiki, I led the development of an in-house online advertising system from scratch, transforming the platform's ability to support sellers in promoting their products effectively. This system was crucial for enhancing visibility for new sellers and improving user experience by delivering relevant advertisements. My role involved designing the technical architecture, building the initial system with a small team, and expanding it to a full-fledged team of 15 engineers, ultimately driving significant revenue growth for the company.

Key Achievements:

• In-House Advertising System: Designed and implemented a robust advertising platform, replacing a third-party solution, which increased control over data and improved ad performance and user experience.
• Revenue Growth: Successfully grew advertising revenue from approximately $750k to nearly $5M annually, making it the fastest-growing business segment within Tiki.
• Team Expansion and Leadership: Expanded the engineering team from 3 to 15 members, establishing clear scopes and responsibilities
• Technical Innovation: Developed advanced features such as real-time ranking and autobidding systems using machine learning, setting a new standard for e-commerce advertising in Vietnam.

At CocCoc, a popular web browser and search engine in Vietnam, I was the Technical Lead for the backend advertisement team. I focused on developing and supporting components that deliver ads across company products, using C++ and Go to enhance the ad-serving infrastructure and optimize ad performance.

Key Achievements:

• SSP Development: Designed and implemented a Supply-Side Platform (SSP) using Go, improving ad delivery efficiency by enabling seamless auctions between CocCoc's internal demand-side platforms and external partners.
• Machine Learning Framework: Developed a framework to integrate machine learning for data collection and model building, significantly improving click-through rate (CTR) predictions.
• CTR Optimization: Increased new tab banner CTR by analyzing user behavioral patterns for more targeted ad placements.
• Automatic Bidding Strategy: Created an automatic bidding strategy to optimize bids within daily budget constraints, reducing manual effort for clients and support teams.
• Ad Placement Expansion: Supported new ad placements, including implementing the Video Ad Serving Template (VAST) format for video ads, ensuring compatibility with third-party products.

At Mail.ru Group, a leading internet company offering a range of services including search engines, payment systems, and social networks, I worked in the department responsible for developing Russia's third-largest web search engine, handling approximately 60 million requests daily. My role involved enhancing various components of the search engine to improve efficiency and user experience.

Key Achievements:

• Data Storage Optimization: Redesigned the data storage system for the plain index, achieving a 30% reduction in size, which improved storage efficiency and performance.
• Aerospike DB Integration: Implemented support for Aerospike DB as a backend for external snippet storage, enabling incremental updates for rapidly changing data such as news.
• Search Result Enhancements: Improved the presentation of data on the search engine results page by developing algorithms for snippet generation and enhancing data completeness, including a Hadoop job for extracting office addresses from web pages.

I contributed to developing a flexible streaming platform, focusing on performance optimization and capability enhancement.

Key Achievements:

• Error Correction: Implemented the RaptorQ algorithm (RFC 6330), enabling UDP use and reducing frame loss in slow networks.
• Video Recognition: Optimized the video recognition server with SSE, achieving a 40% performance boost.
• Memory Management: Added hugepages support, reducing cache misses and accelerating video fingerprint loading, enhancing system efficiency.

At Samsung Electronics, I worked as a Linux System Developer in the Mobile R&D Group, focusing on enhancing system performance for Tizen OS, a Linux-based mobile operating system.

Key Achievements:

• System Daemon Development: Created system daemons like a power manager and network traffic counter, integrating userspace and kernel components for comprehensive statistics.
• Performance Optimization: Improved Tizen OS libraries and applications, notably reducing CPU and power usage by 50% in the camera app through algorithmic enhancements.

At Mail.ru Group, I contributed to the development of Allods Online, focusing on enhancing the game's user interface and text rendering capabilities.

Key Achievements:

• Text Rendering System: Developed a system supporting both LTR and RTL languages, with XML markup for dynamic text styling.
• Widget System Enhancement: Optimized and expanded the internal widget system, adding features and improving performance, with Lua bindings and a WYSIWYG editor for efficient game interface customization.