Built and launched a multi-tenant SaaS control plane where analytics, replay, errors, traces, flags, revenue, people, and agent tooling read from one connected event model.
–Unified product analytics, reliability, releases, experiments, revenue, people, communications, and exports in one console.
–Shipped browser, React, React Native, Node, Python, replay, and native-crash SDKs around one versioned event envelope.
–Exposed permission-filtered product data to coding agents through an MCP server and Management API without bypassing tenant authorization.
Built and launched a multilingual discovery board where free community launches compete on votes, paid placement sits on a separate board ordered by the amount paid, and every visibility claim stays explicit.
–Shipped free daily community launches with sign-in, voting, and review, alongside a paid ranking board, category boards, search, listing pages, and insights.
–Kept paid seats account-free: a completed payment claims a seat and issues a short-lived capability for publishing its listing.
–Enforced price, receipt, publishing, voting, and click-integrity rules in least-privilege Postgres functions rather than trusting the browser.
Built and launched a company-facing platform that turns approved business knowledge into customer agents for websites and messaging channels, with grounded answers, voice, structured actions, and human handoff.
–Shipped one agent workflow across the website widget, Telegram, WhatsApp, Messenger, Instagram, Discord, and KakaoTalk.
–Turned a pasted company URL into a reviewable agent draft, approved knowledge base, embeddable widget, and owner inbox.
–Built production billing, multilingual operation, voice, retrieval, consent controls, and failover around the customer-facing runtime.
Built an agentic AI learning platform that turns source material and learner context into structured courses, lessons, quizzes, audio, video, and AI tutor support across web and mobile.
–Expanded a single course generator into a multi-surface learning system for learners, creators, teams, and mobile users.
–Built AI generation, tutor, RAG, audio, video, and quick-reference workflows around provider-aware execution and learner memory.
–Connected the product to PostgreSQL/Prisma data models, MinIO media storage, Polar billing, and Expo mobile delivery.
Built and evolved Dominion Legist from a legal CMS into a multilingual client platform for understanding services, evaluating the firm, reading legal guidance, and starting a confidential consultation.
–Reframed six legal practice areas as clear, decision-oriented service journeys in Uzbek, English, and Russian.
–Connected the public brand, founder profile, legal publishing, FAQ, and confidential inquiry flow in one platform.
–Kept editorial and media operations manageable through a secure CMS while substantially upgrading the public experience.
Built and launched an Uzbek-first touch-typing academy that treats Uzbek Latin and Cyrillic as first-class keyboard tracks, backed by a server-authoritative scoring engine so results actually mean something.
–Shipped a complete academy: 302+ lessons across five keyboard scripts, a 60-second placement diagnostic, achievements, leaderboards, and three practice games.
–Made Uzbek Latin and Cyrillic first-class typing tracks — including correct oʻ/gʻ apostrophe handling that most typing tools get wrong.
–Built a server-authoritative scoring engine with an anti-cheat replay pipeline, so speeds on the leaderboard are verified, not self-reported.
Built a private insurance agent platform where customers can start an AI consultation, managers can monitor and continue the workflow, and companies can route internal or external phone calls through the same operating console.
–Connected AI-guided customer intake with manager-side follow-up, reservations, call records, and operational review.
–Added internal and external SIP call workflows so companies can move from digital consultation into live advisor conversations.
–Turned insurance guidance into a stateful product flow backed by realtime events, knowledge retrieval, and auditable consultation history.
Built private insurance consultation workflows that let customers talk with an AI assistant, then move into advisor-supported chat, scheduling, and voice follow-up.
–Connected customer self-service consultation with manager-side workflows for review, follow-up, and advisor handoff.
–Added an admin calling surface for SIP-based outbound consultation flows without exposing private customer details.
–Created review surfaces for call history, summaries, and quality checks so the team could monitor AI-assisted conversations.
Built a child-safety reporting product during Seoul Tech Impact Hackathon 2024 and helped the team place second with a mobile-first, privacy-conscious experience.
–Took second place at Seoul Tech Impact Hackathon 2024.
–Designed the product around safety, clarity, and low-friction reporting for vulnerable users.
–Shaped a concept that balanced urgent UX needs with privacy and support-oriented AI assistance.
