Mikhail is available for hire
Hire MikhailMikhail Koviazin
Verified Expert in Engineering
Web Developer
Shanghai, China
Toptal member since September 22, 2021
Bio
Mikhail is a senior full-stack engineer with 12+ years in JavaScript and TypeScript, now focused on AI agents. He builds browser automation with LLM fallbacks, MCP servers, and multi-agent workflows on Claude and Codex, and has authored benchmark tasks for evaluating frontier coding models. Before that, he built scraping systems covering 5+ million products across 1,000+ sites and shipped 50+ browser extensions for Chrome, Firefox, and Safari.
Portfolio
Chinese Marketplaces Integration Gateway
Model Context Protocol (MCP), Claude, Claude Code, Codex...
ProductMappingAI
Claude, Anthropic, Large Language Models (LLMs), LLM Integration, TypeScript...
Cross-Border Shopping Startup
Supabase, React, Vite, PostgreSQL, Web Scraping, Scraping, Data Scraping...
Experience
- Web Development - 12 years
- Web Scraping - 11 years
- Back-end - 11 years
- Node.js - 11 years
- React - 11 years
- JavaScript - 11 years
- Chrome Extensions - 10 years
- TypeScript - 10 years
Work Experience
AI Integration Engineer
2026 - PRESENT
Chinese Marketplaces Integration Gateway
- Built an MCP server with 19 tools for 1688 and Taobao: keyword and image search, full product card extraction, and seller chat.
- Designed seller messaging as a prepare, approve, send flow, with follow-ups guarded by thread versioning so they stop on any new incoming message.
- Built a 24/7 gateway service: a logged-in Chrome under Xvfb in Docker with noVNC for QR login, a SQLite chat mirror with reconciliation, and a reliable send queue.
- Exposed it through an HTTP API with scoped keys and an OpenAPI 3.1 contract, plus a thin MCP server on top with idempotency keys and a retry budget.
- Verified crash recovery after a send with failpoint tests, so a live failure is restored without a duplicate message; 700+ tests across the gateway.
- Recorded architecture decisions as ADRs and covered the MCP server with 267 tests.
- Set up a multi-agent development workflow on Claude and Codex with distinct roles: an orchestrator, implementing agents working in parallel in isolated git worktrees, and a senior reviewer signing off on the plan and every milestone.
Technologies: Model Context Protocol (MCP), Claude, Claude Code, Codex, AI Agent Orchestration, TypeScript, Node.js, Docker, SQLite, Data Extraction, Web Scraping, OpenAPI, API Development, Unit Testing, Chrome DevTools, Linux, Bash, ChatGPT, Orchestration
AI Engineer
2026 - PRESENT
ProductMappingAI
- Built a product data extraction service in Node.js and TypeScript that turns a product page into structured JSON through a three-stage cascade.
- Designed the cascade: declarative per-site YAML mappings first, a generic JSON-LD, OpenGraph and microdata extractor second, an LLM fallback on the Claude API last.
- Routed to the LLM fallback whenever a site had no mapping or its mapping returned data that failed Zod schema validation.
- Set up a multi-agent mapping loop with roles split across models: Claude Sonnet writes site mappings, Claude Opus supervises the flow and prompts, and an independent reviewer judges each PR.
- Added an arbiter step that resolves disputed golden fixtures, so the expected output for each site stays consistent between runs.
- Covered the extractors with about 500 Vitest tests against golden fixtures collected from real product pages.
- Ran the work in branches with a written Definition of Done and recorded design decisions as ADRs.
Technologies: Claude, Anthropic, Large Language Models (LLMs), LLM Integration, TypeScript, Node.js, Web Scraping, Data Extraction, Zod, Vitest, AI Agent Orchestration, Prompt Engineering, Claude Code, Claude API, YAML, Agentic Coding, Playwright, ChatGPT, Orchestration, LLM Agents
Senior Full-stack Engineer (AI Integration)
2026 - 2026
Cross-Border Shopping Startup
- Took over a React and Supabase product generated with an AI app builder and productionized its product-resolution pipeline for US retailer pages.
- Combined direct HTML, Zyte structured and browser-rendered extraction, and Firecrawl, with structured parsing, rate limits, caching, and retries on transient bans.
- Integrated the existing AI size and weight inference into the resolver and fed it extracted product specifications, with Gemini as the main model and GPT-4o-mini as fallback.
- Routed uncertain and hard-blocked results to manual review instead of issuing a final quote, and added provider-level telemetry.
- Fixed production failures in the inherited code, including a client-side enrichment poll that never terminated.
Technologies: Supabase, React, Vite, PostgreSQL, Web Scraping, Scraping, Data Scraping, Website Data Scraping, Web Crawlers, Data Extraction, Data Pipelines, Data Transformation, HTML, HTML DOM, JSON, Proxies, TypeScript, JavaScript, Gemini, OpenAI API, LLM Integration, AI Integration, Large Language Models (LLMs), Prompt Engineering, API Integration, APIs, Firecrawl, Deno, Lovable, ChatGPT, Claude, Back-end
Software Engineer, AI Benchmark Tasks (via Toptal)
2026 - 2026
AI Evaluation Company
- Authored coding tasks for a SWE-Bench-style benchmark built by an AI evaluation company, passed the vetting task, and had a task accepted into the benchmark as verified.
- Turned real defects in large open-source codebases into task statements, reference fixes, hidden tests and graders, checking that the reference fix scored 1.0 and the unmodified code 0.0 before submission.
- Calibrated difficulty by running frontier models against each task and reviewing their traces for where they went wrong and whether the grader judged them fairly.
- Profiled 33 open-source codebases across TypeScript, Go, Rust, and Python in search of benchmark-grade defects.
- Automated the task workflow with shell and Python tooling over Docker-based environments: image builds, deploys, gold and baseline checks, rubric validation.
- Ran the work as long-horizon Claude Code sessions, about 200 of them handed off through state files, with overnight orchestrator runs of up to seven parallel agent threads.
- Contributed a merged fix to the evaluation platform's open-source Python SDK so it imports on systems without Unix sockets.
Technologies: Software Engineering, Artificial Intelligence (AI), Large Language Models (LLMs), Claude Code, AI Agents, Docker, TypeScript, Claude, AI Agent Orchestration, Python, AI Evaluation, Model Evaluation, Benchmarking, Unit Testing, Debugging, Bash, Git, GitHub, Prompt Engineering, Agentic Coding, Open Source, Orchestration, LLM Agents
AI Agent and Browser Automation Engineer (via Toptal)
2026 - 2026
Verifast, Inc
- Rebuilt an inherited RealPage SPA connector, replacing LLM-driven browser steps with deterministic DOM interactions and keeping the LLM as a fallback: runtime went from 7-8 minutes, past the 10-minute limit, to 60-75 seconds at 99%+ reliability.
- Revived a ResMan connector that had been switched off and left broken, with applications processed by hand: every step rebuilt as a deterministic DOM flow with an LLM fallback, and put into production for the first time.
- Built driver-license validation against official state DMV sources with no public API, shipping Florida over plain HTTP with CAPTCHA solving and no headless browser, on a design where adding a new state is a plug-in.
- Implemented I-20 and I-901 (SEVIS fee) form validation through imitated browser requests.
- Generalized the team's reusable workflow template out of two connectors: a step factory with optional browser context, field validation, and skip conditions, plus a DOM-first runner with LLM fallback, and wrote the guide documenting it.
Technologies: JavaScript, TypeScript, AI Agents, OpenAI, Large Language Models (LLMs), Claude Code, Node.js, Playwright, OpenAI API, CAPTCHA, Artificial Intelligence (AI), Agentic Frameworks, SaaS, Cloud Infrastructure, Workflow Automation, Business Process Automation, Software Architecture, AI-assisted Development, LLM Integration, Data Extraction, Data Pipelines, AI Integration, API Integration, Workflows, Integration, Agentic Coding, Mastra, Browserbase, Stagehand, Inngest, Gemini, Automation, Data Transformation, Architecture, System Design, API Management, Prompt Engineering, OpenAI GPT-4 API, JSON, Communication, System Architecture, Agentic Workflow Design, Systems Design, AI Agent Orchestration, Agentic AI, Scripting, Automations, AI Development, AI Automation, n8n, AI Architecture, AI Engineering, Jira, ChatGPT, Claude, Orchestration, LLM Agents, Back-end
Senior Chrome Extension Engineer (via Toptal)
2024 - 2026
Saleo Inc.
- Developed and maintained the browser extension for an AI-powered live demo platform that personalizes and controls demo data directly inside live third-party products.
- Implemented the Demo Sync feature that significantly decreased time costs for enterprise sales, presales, and enablement teams.
- Integrated the existing real-time data sync into personalized demo charts and tables, substituting API and WebSocket data so child demos override only the data they need.
- Delivered further full-stack features across the extension, portal, and back end: external start-demo URLs, personalized global CSS, and URL matching enhancements.
- Fixed performance in the React and Redux portal and extension UI: removed expensive deep copies of demo data, loaded demos in parts instead of all at once, and sped up slow demo search and the URL list history view.
Technologies: React, Chrome Extensions, Node.js, Amazon Web Services (AWS), AWS Lambda, WebStorm, JavaScript, TypeScript, Git, Web App Automation, Web Development, Web Extensions, APIs, Full-stack Development, AI Tools, API Integration, Vibe Coding, User Interface (UI), API Development, Authentication, Data Engineering, WebSockets, Claude Code, Cloud Infrastructure, Workflow Automation, Business Process Automation, Software Architecture, AI-assisted Development, Data Extraction, Data Pipelines, Workflows, Integration, Agentic Coding, Prompt Engineering, System Architecture, Systems Design, Storybook, Jira, DevOps, Back-end, Terraform
Senior Browser Extension Developer (via Toptal)
2024 - 2024
Numen Capital LLP
- Rewrote the extension prototype from MV2 to MV3 (to be able to publish the extension in the Chrome Web Store). Refactored state management system and text highlighting algorithms.
- Implemented PDF file interception in the extension (local and web-based documents) under MV3, with text highlighting in a custom PDF viewer (PDF.js).
- Rendered all on-page UI inside Shadow DOM so host-page styles cannot reach it, plus a keep-on-top routine that re-checks its own stacking against other extensions.
- Built new extension UI components and fixed issues in the existing prototype UI.
- Raised a privacy and performance problem the client had not asked about: the whole page HTML was leaving the browser on every user action.
Technologies: JavaScript, React, Chrome Extensions, Full-stack, MV3, MV2, Application State Management, PDF.js, WebStorm, TypeScript, Git, Web Development, Web Extensions, APIs, Full-stack Development, AI Tools, API Integration, User Interface (UI), Document Processing, Data Engineering, Artificial Intelligence (AI), Agentic Frameworks, Cloud Infrastructure, LLM Integration, AI Integration, Integration, Agentic Coding, Fintech, Financials
Browser Extension Deployment Specialist (via Toptal)
2024 - 2024
Stage Front Tickets, Inc.
- Reviewed a ticket broker's internal pricing extension for Chrome Web Store eligibility and stripped every permission the code did not actually use.
- Wrote the Permission Justification, Single Purpose and Data Usage sections of the submission, which is where store reviews usually fail.
- Prepared the store submission and handed it back to the client, and registered and handed over their Chrome Web Store developer account.
- Advised on the policy questions behind it: privacy-policy updates after publication, and remote or dynamically evaluated code as a hard blocker.
Technologies: Google Chrome, Chrome Extensions, JavaScript, Manifest V3, TypeScript, WebStorm, React, Git, Web Development, Web Extensions, APIs, API Integration, Code Auditing, Payment APIs, Integration, Payments
Senior Browser Extension Developer (via Toptal)
2022 - 2024
Shady.Club Ltd
- Took a brand-ethics labeling extension forward from MVP, surfacing Good On You sustainability ratings on product cards across European fashion retail sites.
- Rebuilt the brand-matching system between product cards and the brand database, eliminating false positives and negatives for brands with overlapping names (deterministic, no AI).
- Improved the database architecture to match brand data faster.
- Reworked the on-page UI and pop-ups to run smoothly over heavy third-party SPA sites.
- Shipped in-product AI: similar-product search by image, where a click on any product picture returns higher-rated alternatives.
Technologies: JavaScript, Chrome Extensions, APIs, Browser Plugins, WebStorm, React, TypeScript, Git, Web Development, Web Extensions, Full-stack Development, AI Tools, API Integration, User Interface (UI), Data Engineering, Artificial Intelligence (AI), Cloud Infrastructure, LLM Integration, AI Integration, Integration, Agentic Coding, Minimum Viable Product (MVP), MVP Design
Senior Full-stack Developer (via Toptal)
2023 - 2023
Kenneth Corigliano
- Developed a full-stack real-estate product: a browser extension for sharing property notes across listing sites, plus the back end and an admin dashboard to manage users and notes.
- Detected listing pages on three major real-estate sites, parsed the address off the page and keyed notes to it, with a manual link-up when the addresses did not match.
- Implemented a user-friendly notes-sharing flow for users without an installed browser extension, which allows those users to install the extension and review the property notes.
Technologies: Browser Plugins, Chrome Extensions, JavaScript, Full-stack, Azure, MV3, React, WebStorm, TypeScript, Git, Web Development, Web Extensions, APIs, Full-stack Development, API Integration, User Interface (UI), Integration, Database Architecture
Senior Full-stack Web Developer
2019 - 2023
Fatcoupon Technology
- Architected and developed a cross-browser extension for auto-applying coupon codes and cashback across 1,000+ online stores, reaching 30,000+ users at a 4.8/5 rating.
- Created a system for automated web scraping and price updating over 5,000,000 products and deals from over 1,000 websites.
- Automated routine business tasks such as scraping deals and coupons, registering accounts, and cleaning affiliate links.
- Led a team of three software engineers, mentoring and reviewing the code.
- Worked closely with the product owner and product manager on analyzing competitors, business needs, and development vectors from the technical side.
- Built the React, Redux, and TypeScript dashboard of the price-monitoring system and the internal back-office screens, with no designer, agreeing with the business on what each screen had to show.
Technologies: Node.js, React, Chrome Extensions, Web Scraping, Web App Automation, JavaScript, MongoDB, Redux, TypeScript, Web Extensions, REST APIs, HTML, CSS, Firefox Extensions, Software Engineering, WebStorm, Git, Webpack, Web Development, Web Crawlers, Large-scale Web Crawlers, GitHub, Figma, Amazon S3 (AWS S3), ETL, Scraping, Amazon Web Services (AWS), Data Scraping, Axios, APIs, REST, Databases, Robotic Process Automation (RPA), Browser Plugins, Full-stack, Apify SDK, Browser Automation, HTML DOM, Manifest V3, Google Chrome, Workflow Automation & System Integration, Full-stack Development, API Integration, Next.js, User Interface (UI), API Development, Authentication, Technical Architecture, Playwright, Data Analysis, Data Engineering, Data Mining, Artificial Intelligence (AI), Payment APIs, SaaS, Cloud Infrastructure, Workflow Automation, Business Process Automation, Software Architecture, Data Extraction, Data Pipelines, Workflows, Integration, Docker, Automation, Website Data Scraping, ETL Pipelines, ETL Tools, Data Transformation, Architecture, System Design, Solution Architecture, Data Reporting, API Management, JSON, Proxies, Communication, Minimum Viable Product (MVP), System Architecture, Technical Leadership, Systems Design, Leadership, Payments, Scripting, Automations, NoSQL, Database Architecture, CAPTCHA, Product Management, Product Strategy, Jira, DevOps, Back-end
Senior Full-stack Web Developer (via Toptal)
2022 - 2022
WFHLAND Inc.
- Architected and developed a browser extension for making on-page web notes across websites and on online/local PDF files.
- Integrated an open-source web notes engine with a browser extension and improved its algorithms for better (de)serialization of web notes over various websites, including complicated cases with highly dynamic websites and online/local PDF files.
- Integrated an open-source library for rendering PDF files and improved some of its features. Developed web notes functionality for PDF files.
Technologies: TypeScript, JavaScript, React, Chrome, Front-end Build Tools, Chrome Extensions, HTML Canvas, Front-end, Front-end Development, WebStorm, Git, Web Development, Web Extensions, APIs, Full-stack Development, API Integration, User Interface (UI), Document Processing, Integration
Senior Browser Extension Developer (via Toptal)
2021 - 2022
1Blocker LLC
- Adapted open-source ad-blocking scriptlets to Safari, where blocking runs through content blocker rules and page scriptlets rather than request interception, and wrote the desktop- and mobile-specific ones the open-source sets lacked.
- Extended and deeply refactored the existing ad-blocking web extension functionality for Safari on iOS and macOS.
- Developed a custom solution for data storage that works in a specific web extension environment.
- Reverse-engineered competitor ad-blocking extensions (open-source and proprietary) to inform architecture decisions, and produced a feature comparison for the client.
- Wrote targeted blocking logic for YouTube and Twitter to counter aggressive anti-ad-block measures, which change on the sites' side constantly.
Technologies: Web Extensions, iOS Extensions, Safari Extensions, TypeScript, Webpack, Jest, ESLint, Lint, Testing, HTML, CSS, Software Engineering, WebStorm, JavaScript, Git, Web Development, GitHub, Browser Plugins, Full-stack, Web App Automation, Browser Automation, HTML DOM, Manifest V3, Google Chrome, MacBook, MacOS, Chrome Extensions, API Integration, User Interface (UI), Code Auditing, Software Architecture, Integration, System Architecture, Systems Design, Unit Testing
Full-stack Web/Browser Extension Developer
2016 - 2020
Freelance Clients
- Developed and maintained over 50 browser extensions and web service projects.
- Used web scraping to extract data from websites, including Amazon, Google, Facebook, Instagram, Twitter, LinkedIn, and others.
- Built LinkedIn and Sales Navigator automation for lead generation and enrichment: exporting profile and connection data (name, email, company, title, location) to spreadsheets and tagging company accounts from a source list.
- Created web automation across different sites: checkout automation, bulk messaging, and personalized outreach flows built from each profile's own data.
- Automated full authenticated user flows on heavily bot-protected platforms (login, navigation, search, posting, data collection), including IP rotation, proxy, and browser fingerprint management.
- Extracted billing for a subscription product into a standalone service on a local payment provider: invoices, direct debit, subscriptions, and installment plans, with an API the client's other sites could plug into.
- Integrated two payment providers for a charity foundation, including recurring monthly donations, and built the reporting pipeline that ingested its whole donation history of about 13,800 records.
Technologies: Node.js, React, Chrome Extensions, MongoDB, Bootstrap, TypeScript, Amazon Web Services (AWS), Git, Redux, Web Extensions, APIs, Software Development, HTML, CSS, Firefox Extensions, Software Engineering, WebStorm, Web Scraping, JavaScript, Webpack, Puppeteer, CheerioJS, Web Development, Web Crawlers, Large-scale Web Crawlers, GitHub, Slack, Figma, Amazon S3 (AWS S3), ETL, Scraping, Data Scraping, Axios, REST, Databases, Robotic Process Automation (RPA), Browser Plugins, Full-stack, Apify SDK, Web App Automation, Browser Automation, HTML DOM, Manifest V3, Google Chrome, Workflow Automation & System Integration, Zapier, AI Agents, Large Language Models (LLMs), OpenAI, Python, Full-stack Development, Stripe, Stripe API, API Integration, Scrapy, Next.js, User Interface (UI), API Development, Authentication, Document Processing, Technical Architecture, Playwright, PostgreSQL, Redis, LinkedIn, Data Engineering, Vercel, Prisma, Code Auditing, Payment APIs, SaaS, Cloud Infrastructure, Facebook, Workflow Automation, Business Process Automation, Lead Generation, LinkedIn Sales Navigator, Software Architecture, Data Extraction, Data Pipelines, Workflows, Integration, Docker, SQL, Automation, Website Data Scraping, ETL Pipelines, ETL Tools, OpenAI GPT-4 API, Google API, JSON, Proxies, Selenium, System Architecture, Technical Leadership, Systems Design, Leadership, Payments, Unit Testing, Scripting, Automations, PhantomJS, Spintax Generator, Material UI, Microsoft Excel, Google Sheets, CSV, NoSQL, CAPTCHA, Jira
Software Developer
2014 - 2016
66Bit
- Developed a web service for aggregation and one-point management of four context ad networks and analytics services.
- Maintained and added new features for the English learning desktop application.
- Led a team of two developers, reviewing and mentoring them while working on English learning mobile and desktop apps.
- Helped develop a SaaS for obtaining access to wifi access points through social networks.
Technologies: React, ASP.NET MVC, C#, JavaScript, Entity Framework, Windows Presentation Foundation (WPF), Windows Forms (WinForms), PHP, .NET, Git, APIs, Software Development, HTML, CSS, Software Engineering, WebStorm, Web Development, GitHub, ETL, Scraping, Data Scraping, REST, Databases, Full-stack, HTML DOM, Workflow Automation & System Integration, Full-stack Development, API Integration, User Interface (UI), Software Architecture, Data Extraction, Data Pipelines, Integration, SQL, Architecture, System Design, Solution Architecture, API Management, Google API, System Architecture, Systems Design, MySQL, CSV, Razor Pages, Back-end
Project History
FatCoupon Browser Extension
https://fatcoupon.com/chrome-coupon-extensionA cross-browser extension for auto-applying coupons and obtaining cashback on 1,000+ online stores, carried from the first MVP to a production product with 30,000+ users at a 4.8/5 rating.
• Researched competitors and wrote the software requirement specifications.
• Planned the architecture of the project and chose the most appropriate technology stack.
• Architected a uniform JavaScript solution for auto-applying coupons on 1,000+ websites in multiple desktop browsers and mobile apps (Android, iOS) via WebView.
• Kept server-delivered store connectors working after Manifest V3 banned remotely delivered code: connectors are parsed into an AST and run by an interpreter inside the extension sandbox, a subsystem still present in the published build.
• Developed the extension's front end using React, Redux, and TypeScript.
• Integrated an auto-applying coupons function into mobile apps using Kotlin and Swift.
• Led a team of three developers while integrating 1,000+ websites with extensions.
• Researched competitors and wrote the software requirement specifications.
• Planned the architecture of the project and chose the most appropriate technology stack.
• Architected a uniform JavaScript solution for auto-applying coupons on 1,000+ websites in multiple desktop browsers and mobile apps (Android, iOS) via WebView.
• Kept server-delivered store connectors working after Manifest V3 banned remotely delivered code: connectors are parsed into an AST and run by an interpreter inside the extension sandbox, a subsystem still present in the published build.
• Developed the extension's front end using React, Redux, and TypeScript.
• Integrated an auto-applying coupons function into mobile apps using Kotlin and Swift.
• Led a team of three developers while integrating 1,000+ websites with extensions.
Automated Product Scraping System
Automated web scraping and price alerting across 1,000+ websites, covering 5,000,000+ products and deals per day at peak.
• Researched competitors and wrote the software requirement specifications.
• Planned the project architecture and chose the technology stack.
• Architected a uniform JavaScript solution for scraping 1,000+ websites, including those requiring browser rendering and scraping-protection bypass.
• Built the crawling layer on Apify SDK v1, the codebase that later became Crawlee, running it locally rather than on the hosted platform, with custom scraper base classes over its Cheerio and Puppeteer crawlers.
• Developed the dashboard front end using React, Redux, and TypeScript, and the back end using Node.js, Express, Passport, MongoDB, and TypeScript.
• Set up the CI/CD pipeline and ran the system on AWS EC2 under PM2, with the same team of three integrating new sites.
• Researched competitors and wrote the software requirement specifications.
• Planned the project architecture and chose the technology stack.
• Architected a uniform JavaScript solution for scraping 1,000+ websites, including those requiring browser rendering and scraping-protection bypass.
• Built the crawling layer on Apify SDK v1, the codebase that later became Crawlee, running it locally rather than on the hosted platform, with custom scraper base classes over its Cheerio and Puppeteer crawlers.
• Developed the dashboard front end using React, Redux, and TypeScript, and the back end using Node.js, Express, Passport, MongoDB, and TypeScript.
• Set up the CI/CD pipeline and ran the system on AWS EC2 under PM2, with the same team of three integrating new sites.
Facebook Dating Extension and Website
A subscription dating product built around Facebook groups: a cross-browser extension that added its own matching controls to Facebook pages, the web application behind it, and a separate billing service.
• Developed the front end for both the extension and the website using React and Bootstrap.
• Developed the back end using Node.js, Express, MongoDB, and Webpack.
• Built subscription billing as a separate service with its own payment-gateway integration, so subscriptions and in-app purchases were handled independently of the main application.
• Set up the CI/CD pipeline for deploying on Heroku.
• Developed the front end for both the extension and the website using React and Bootstrap.
• Developed the back end using Node.js, Express, MongoDB, and Webpack.
• Built subscription billing as a separate service with its own payment-gateway integration, so subscriptions and in-app purchases were handled independently of the main application.
• Set up the CI/CD pipeline for deploying on Heroku.
Contextual Ad Network Management Service
Data aggregation and one-point management web service for four contextual ad networks and analytics services.
• Wrote the technical specification, chose the stack, and built the first working prototype alone before the project got a repository and a team.
• Integrated and unified management functions of four different APIs: Google Ads, Google Analytics, Yandex Direct, and Yandex Metrica.
• Designed one domain model over both ad platforms with two-way synchronization, so changes made in the service propagated out into the ad networks and network-side state came back in.
• Designed bidding automation around the platform's rate limit (3,000 changes per campaign per day, so 125 keywords on an hourly cycle), with full caching required to stay inside the quota.
• Built the server-rendered UI for most of the product screens (ASP.NET MVC Razor, Bootstrap, jQuery), including ad editing and the on-site form and click collector; a dedicated front-end developer carried the layout and most client-side JavaScript.
• Built the back end using ASP.NET MVC, EF6, MS SQL Server, and Windows IIS Server; the service ran in production for a year against a real advertiser's live campaigns.
• Wrote the technical specification, chose the stack, and built the first working prototype alone before the project got a repository and a team.
• Integrated and unified management functions of four different APIs: Google Ads, Google Analytics, Yandex Direct, and Yandex Metrica.
• Designed one domain model over both ad platforms with two-way synchronization, so changes made in the service propagated out into the ad networks and network-side state came back in.
• Designed bidding automation around the platform's rate limit (3,000 changes per campaign per day, so 125 keywords on an hourly cycle), with full caching required to stay inside the quota.
• Built the server-rendered UI for most of the product screens (ASP.NET MVC Razor, Bootstrap, jQuery), including ad editing and the on-site form and click collector; a dedicated front-end developer carried the layout and most client-side JavaScript.
• Built the back end using ASP.NET MVC, EF6, MS SQL Server, and Windows IIS Server; the service ran in production for a year against a real advertiser's live campaigns.
Used-book Arbitrage Monitoring System
A market-monitoring system for a used-book arbitrage business: it watches the client's ISBN list across book marketplaces and alerts within seconds on new listings and price drops. Running since 2019, three generations of the codebase, still in production.
• Built the whole pipeline on plain HTTP with axios and Cheerio, no browser anywhere, so a sweep costs milliseconds and fits inside the client's proxy budget of 1,000,000 requests per month.
• Rewrote the system in TypeScript and moved it into Docker on the client's AWS EC2, with per-second ticks and a schedule that polls faster during peak buying hours.
• Replaced HTML scraping of a 50-million-record catalog with batched queries against the site's own search index: 15,000 watched ISBNs are now covered by three requests per pass.
• Made silent failure visible: a sweep that parses zero listings raises a warning instead of looking like a quiet market, which is how a markup change had gone unnoticed for days.
• Built the whole pipeline on plain HTTP with axios and Cheerio, no browser anywhere, so a sweep costs milliseconds and fits inside the client's proxy budget of 1,000,000 requests per month.
• Rewrote the system in TypeScript and moved it into Docker on the client's AWS EC2, with per-second ticks and a schedule that polls faster during peak buying hours.
• Replaced HTML scraping of a 50-million-record catalog with batched queries against the site's own search index: 15,000 watched ISBNs are now covered by three requests per pass.
• Made silent failure visible: a sweep that parses zero listings raises a warning instead of looking like a quiet market, which is how a markup change had gone unnoticed for days.
Marketplace Product Research System
A product research pipeline for a sourcing company in Yiwu, China: a browser extension and an Excel automation layer that pulled listing, pricing, and real sales data off Amazon and eBay into a research workbook. Over five months on site, it carried 127 product studies, covering about 3,400 competitor listings and 2,400 product photographs on a fixed 13-field schema.
• Built a Chrome extension (Manifest V2) that extracted listing data from Amazon and eBay into the spreadsheet, plus an Excel VBA layer that fitted product photos into cells by ASIN and generated per-row link buttons.
• Replaced a paid tool's estimated monthly sales with real per-date transaction counts parsed from eBay's Purchase History, with captcha detection and request throttling built in from the start.
• Caught the paid estimator returning wrong numbers for a whole category two weeks in, reported it in writing rather than shipping the figures, and switched the company onto another tool the same day.
• Ran the marketplace and payment accounts behind the research, and rewrote supplier copy into sellable listings. At least one study went from research to a sourcing decision and factory samples.
• Built a Chrome extension (Manifest V2) that extracted listing data from Amazon and eBay into the spreadsheet, plus an Excel VBA layer that fitted product photos into cells by ASIN and generated per-row link buttons.
• Replaced a paid tool's estimated monthly sales with real per-date transaction counts parsed from eBay's Purchase History, with captcha detection and request throttling built in from the start.
• Caught the paid estimator returning wrong numbers for a whole category two weeks in, reported it in writing rather than shipping the figures, and switched the company onto another tool the same day.
• Ran the marketplace and payment accounts behind the research, and rewrote supplier copy into sellable listings. At least one study went from research to a sourcing decision and factory samples.
Education
2017 - 2020
Master's Degree in Software Engineering
Harbin Institute of Technology - Harbin, China
2012 - 2016
Bachelor's Degree in Applied Informatics
Ural State University of Economics - Yekaterinburg, Russia
Skills
Libraries/APIs
Node.js, React, Puppeteer, CheerioJS, REST APIs, API Development, Playwright, Claude API, iOS Extensions, Stripe, Stripe API, OpenAI API, Google API, Zod, Entity Framework, Windows Forms (WinForms), PhantomJS, Google Analytics API, Google Ads API, OpenAPI
Tools
WebStorm, Git, Webpack, Claude Code, Claude, GitHub, Front-end Build Tools, ChatGPT, Microsoft Excel, Google Sheets, Codex, Vitest, WebView, SourceTree, Slack, Figma, Zapier, Prisma, n8n, Jira, Terraform
Languages
JavaScript, TypeScript, HTML, CSS, C#, SQL, Excel VBA, PHP, Kotlin, Swift, Python, Bash, YAML
Frameworks
Redux, Express.js, Chrome, Bootstrap, .NET, Jest, Scrapy, Next.js, Agentic Frameworks, ASP.NET MVC, Windows Presentation Foundation (WPF), Material UI, Selenium, Razor Pages
Paradigms
ETL, REST, HTML DOM, Automation, Model Context Protocol (MCP), Testing, Unit Testing, DevOps
Platforms
MacOS, Google Chrome, Amazon EC2, Amazon Web Services (AWS), Azure, AWS Lambda, Vercel, Docker, Linux, Heroku, Windows Server, Windows, PDF.js
Storage
Data Pipelines, JSON, MongoDB, Amazon S3 (AWS S3), Databases, PostgreSQL, NoSQL, Database Architecture, SQLite, MySQL, Redis
Other
Web Scraping, Chrome Extensions, Software Development, Web App Automation, Apify SDK, Facebook, Web Development, Web Extensions, APIs, Software Engineering, Firefox Extensions, Web Crawlers, Large-scale Web Crawlers, Scraping, Data Scraping, Robotic Process Automation (RPA), Browser Plugins, Full-stack, Browser Automation, Manifest V3, Front-end, Front-end Development, MV3, MV2, Workflow Automation & System Integration, MacBook, AI Agents, Large Language Models (LLMs), OpenAI, Full-stack Development, API Integration, Agentic AI, AI Agent Orchestration, AI Integration, Artificial Intelligence (AI), Technical Architecture, Data Engineering, CAPTCHA, Code Auditing, Anthropic, Workflow Automation, Business Process Automation, Software Architecture, AI-assisted Development, LLM Integration, Data Extraction, Workflows, Integration, Agentic Coding, ETL Pipelines, Architecture, Website Data Scraping, Data Transformation, Proxies, System Architecture, Systems Design, Scripting, Automations, AI Automation, AI Architecture, AI Engineering, Debugging, Chrome DevTools, Payment Gateways, Safari Extensions, Axios, Algorithms, Application State Management, AI Tools, AI Development, Vibe Coding, User Interface (UI), Authentication, Document Processing, Data Analysis, Data Mining, LinkedIn, Payment APIs, SaaS, Cloud Infrastructure, Lead Generation, LinkedIn Sales Navigator, ETL Tools, System Design, Solution Architecture, Data Reporting, API Management, Prompt Engineering, OpenAI GPT-4 API, Communication, Minimum Viable Product (MVP), MVP Design, Technical Leadership, Agentic Workflow Design, Leadership, Payments, CSV, Storybook, Product Management, Product Strategy, AI Evaluation, Model Evaluation, Benchmarking, Open Source, Orchestration, LLM Agents, Back-end, CI/CD Pipelines, Spintax Generator, Yandex.Direct, Yandex.Metrika, ESLint, Lint, HTML Canvas, WebSockets, Supabase, Mastra, Browserbase, Stagehand, Inngest, Gemini, Supply Chain, Supply Chain Operations, Vite, Firecrawl, Deno, Lovable, Fintech, Financials
Collaboration That Works
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
1
Share your needs
Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2
Choose your talent
Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3
Start your risk-free talent trial
Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.
Top talent is in high demand.
Start hiring