Oleksandr (Alex) Bodrov

Own GiveStar’s agentic / AI-assisted SDLC: designed the orchestration agent, cross-repo knowledge base, autonomous ticket-to-merge-request pipeline, and measurement framework across a brownfield estate (.NET, Next.js, React Native, GitLab CI, Linear) with ~14 engineers.

• Designed and built GiveStar’s AI-assisted SDLC (orchestration agent, cross-repo knowledge base, ticket-to-merge-request pipeline, and measurement framework) across a 14-engineer brownfield estate; merged code MRs per engineer rose 4.6 to 5.8 per week (+26%) with no headcount change, and team merged MRs rose +43%/wk (Linear story points +27%/wk).
• Cut development-stage cycle time 40% median (In Progress to code complete 3.3 to 2.0 days; p85 18 to 10 days) and ticket pickup time 54% (To Do to In Progress 4.6 to 2.1 days); QA on human tickets fell 2.8 to 1.3 days.
• Shipped AI-authored merge requests at 90% acceptance (Jun–Aug 2026) with 0% merging without review comments, while holding end-to-end lead time flat at ~14 days and MR-created-to-merged roughly steady (0.64 to 0.70 d median).
• Moved quality signals the right way under the same rollout: hotfix MRs -72%, review bounce-backs 33% to 27%, bug share of completed work 9% to 8%.
• Diagnosed that the 40% development-stage gain was absorbed by prioritisation and review/release queues (customer lead time unchanged) and framed that as the next system bottleneck rather than a finished win.
• Architected governance for an LLM-maintained cross-repository knowledge base with 130+ curated pages (repository routing, bounded handoffs, automated review rounds, human-takeover detection, allow-listed tools, ticket eligibility).
• Led a 35+ component React/Next.js design-system rollout to standardize product delivery across frontend repositories.
• Raised unit-test coverage from 0% to about 35% across GiveStar’s frontend applications; built Playwright E2E from scratch on GitLab CI for two projects, covering critical user paths.

Stack .NET · Next.js · React Native · TypeScript · Claude / Anthropic API · GitLab CI/CD · Linear · Playwright · PostHog · Sentry

Owned production MCP integrations and shared-platform reliability: Python/FastAPI on Modal, OAuth hardening, multi-tenant security.

• Contributed across about 25 MCP integrations; owned seven end-to-end: Shopify, Meta Ads, Triple Whale, Modash, Jira, OpenRouter, and a financial API integration.
• Built Python/FastAPI MCP servers on Modal with OAuth 2.0 or API-key auth; supported a broad integration surface including Hugging Face and Google Drive.
• Fixed shared OAuth retry infrastructure: 72% fewer total MCP errors and 73% fewer dead-token failures in a measured 24-hour window.
• Closed an over-broad default-permission gap in a multi-tenant financial integration; built a reusable YAML-driven MCP simulation/test harness.

Stack Python · FastAPI · Modal · OAuth 2.0 · MCP · OpenRouter · Hugging Face

Founded, designed, built, launched, and operate three products/businesses end-to-end: from system architecture and implementation to users and day-to-day operations. Gathered and mentored a small group of contributors along the way; several went on to stronger engineering roles.

• OhMyDesk (AI-native coworking ops): shipped to about 30 workspace operators and 100 members; multilingual AI Operator Autopilot (Gemini, Supabase Edge Functions, Postgres RLS, allow-listed tools) with propose → confirm → execute → audit, grounded outputs, confidence and knowledge-gap logging, and anti-injection controls.
• Trudify: built and operate a multilingual services marketplace end-to-end with Next.js, DeepL, Stripe, Google Maps, and Telegram automation.
• CodeBurg: founded and operate a technology-focused coworking business and its supporting digital product and operational systems.

Stack Gemini · Supabase · PostgreSQL / RLS · Edge Functions · TypeScript / React · Next.js · Stripe · DeepL