Apr 2025 – Feb 2026 · 10 mos · Belgrade, Serbia · Hybrid

• Developed the Martian Python SDK for AI Safety Hackathon participants, providing seamless access to LLM providers and empowering them to train custom routers optimized for their target quality metrics
• Benchmarked and analyzed a range of LLMs - assessing performance, cost efficiency, and capability coverage - to inform and refine routing decisions.
• Contributed to product feature ideation and user-story mapping, driving cross-functional alignment on roadmap priorities
• 🏆 Awarded Great Communicator - recognized for clear, effective, and impactful communication across teams and stakeholders.

Feb 2024 – Apr 2025 · 1 yr 2 mos · Hybrid

• Word Web Canvas team

Nov 2022 – Feb 2024 · 1 yr 3 mos · Hybrid

• Bing's Crawler team.
• Supported and enhanced the service responsible for selecting web pages for crawling.

Apr 2021 – Oct 2022 · 1 yr 6 mos · Moscow, Russia

• Roles and responsibilities --
• I design and implement algorithms of warehouse replenishment. I design hypotheses which could improve stock availability for each of the warehouses. I test those hypotheses and implement the successful ones for production usage by the replenishment team.
• I optimize the assortment of goods for the entire marketplace.
• I intensively work with large sets of data using SQL, Spark, Pandas, and use our internal map-reduce calculation system. I’m highly proficient with SQL (YQL dialect) which is reflected in my internal Top Performer rating. I’m designing and building dashboards and charts in Tableau and Datalens which support decision making by our business users.
• I interact a lot with the business users and other teams within the company, so I have a good experience of how to establish effective contacts and communications to achieve the project goals.
• Achievements --
• Increased stock availability by 5 pp by implementing a new model of safety stock calculation, which led to sales increase by 2 pp
• Improved unit economics of the marketplace by 3 pp by optimizing the structure of goods storage using a developed model of sales and costs
• Increased the rate of automatically placed replenishment orders by 4 pp by implementing a simulated annealing algorithm
• Improved the warehouse performance by enhancing the replenishment optimization engine with the information on weights and sizes of goods that fit the warehouse receiving capacity
• Developed a long-term forecast of warehouse utilization that depends on growing variety of goods and growing demand, which led to a management decision of renting out a part of the warehouse
• About the marketplace --
• Yandex.Market is one of the largest online marketplaces in Russia. It includes seven big warehouses (not counting the helper ones) across the country, managing more than 16 million of SKUs (FMCG, DYI, Electronics, etc.). The warehouses are shipping more than 100,000 orders daily.

Nov 2018 – Apr 2021 · 2 yrs 5 mos · Moscow, Russia

• Roles and responsibilities --
• At Y.Vzglyad (Y.”Look”), I joined the team as a software engineer, but the team work priorities quickly made me responsible for all projects related to data analysis.
• I built a data warehouse, created business dashboards and reports for the service metrics. I was designing and testing product hypotheses and doing ad-hoc research projects.
• I worked with Big data and MapReduce systems (Spark over Hadoop like) and build ETL piplines (Apach Airflow like)
• On a daily basis, I was writing a lot of complex SQL queries (YQL, Clickhouse, PostgreSQL) and python code (pandas, numpy, requests). From time to time, I was also involved as a software engineer in backend development (python+flask, node.js+express).
• Achievements --
• Organized a data warehouse that provides the structured data for building analytical reports, dashboards with KPI metrics
• Improved the measurement quality of brand lift campaign effectiveness by 5 pp
• Developed a quality assessment benchmark for of video creatives
• Automatized user matching and collecting survey forms by implementing abstract API, which sped up
• getting survey by 2 times
• Built survey standard costing model with different axis, which led to reducing costs by 20pp
• About company --
• Y.Vzglyad is a comprehensive tool for marketing research. It allows users to create survey forms, select the target audience (age range, region, etc.) that will be automatically collected within the Yandex ecosystem, run the research, and analyze the visualized results.

Apr 2016 – Nov 2018 · 2 yrs 7 mos · Yekaterinburg, Sverdlovsk, Russia

• Merkata was a startup company building a SaaS point-of-sale (POS) system fully compliant with complex government regulations.
• Roles and responsibilities --
• I was a .NET backend developer and implemented a comprehensive business logic for Merkata POS.
• I helped the team elaborate business requirements and was designing technical solutions that met these requirements and were relatively quick to develop.
• I implemented the POS back office web interface (.NET Web API, IIS, RabbitMQ, MongoDB, Vue.js) which allowed users to manage chains of stores and analyze their sales data.
• I was also involved in development of the POS terminal application based on the Xamarin framework for both Android and Windows.
• I was taking part in release management and was involved in client migrations to newer versions of the product.
• Achievements --
• At some moment of time, the previous front end tech lead left the company, I stepped in and conducted evaluation and selection of the front end UI framework, so I recommended Vue.js and we moved to that framework.
• Based on the business requirement specifications, I designed and implemented the business logic of an integration layer that exchanged alcohol tracking data with an external government-owned information system (EGAIS).
• I built a subscription management system for Merkata POS. I applied graph algorithms to code a high-performing calculation of overlapping subscriptions based on their start dates and due dates.
• I implemented data synchronization between the POS terminal and the backend server.
• I implemented integration testing using AutoFixture.
• From time to time, I was doing support duties for our service, identifying bugs and fixing them in different parts of the system.
• Soon I became a team leader for a group of cross-functional engineers consisting of 4 people.

Oct 2014 – Apr 2016 · 1 yr 6 mos · Yekaterinburg, Sverdlovsk, Russia

• CRM system design and development for the insurance broker company using MVC, .NET Web API + Angular JS.
• Integration with online payment gateways.
• Design and development of a brand-new insurance calculator (both the backend and frontend), which improved the overall customer experience and increased the lead conversion rate by 20%.
• Reports automation from scratch.