Mykhailo (Mike) Severa
Principal Product Engineer • Entrepreneur
michael.svr@gmail.com · linkedin.com/in/msevera · github.com/msevera · msevera.com
Summary
Principal Product Engineer with 15+ years building and shipping web and mobile products, now focused on AI-native development — multi-agent systems, RAG, vector search, and real-time voice, backed by a spec-driven, agentic engineering workflow. Former Co-founder and CTO at a Y Combinator and a16z crypto-backed company.
Skills
Experience
Principal Product Engineer, AI LeadershipFeb 2026 – Present
- Architected and led a team of 3 product engineers building an AI-native leadership platform on a NestJS + GraphQL + PostgreSQL/TypeORM backend with a Next.js (App Router) frontend.
- Designed a multi-agent system of 9 specialized agents (interview, aggregation, follow-up, chat, weekly check-in) on the Vercel AI SDK, with Braintrust for prompt management and evaluation tracing.
- Built RAG over a Turbopuffer vector store with LlamaIndex ingestion, mem0 long-term memory, and ElevenLabs real-time voice (STT/TTS).
- Implemented real-time UX with GraphQL subscriptions over Redis, plus better-auth / Google OAuth, anonymous-to-user account migration, encryption at rest, and security hardening (Helmet, GraphQL depth/complexity limits).
- Established a spec-driven, agentic engineering practice — shipped 120+ well-scoped features (OpenSpec change proposals) with portable agent tooling across Cursor, Claude Code, and Codex, plus codified conventions and anti-pattern enforcement to keep AI-generated code production-grade.
- Backed it with engineering rigor: Sentry observability across API and web, semantic-release, and a Vitest + Testcontainers test suite.
Independent Ventures2025 – 2026
AI products I designed and built end-to-end as an entrepreneur.
- Gogue AI
Designed and built an AI-powered learning app that delivers personalized audio lectures based on content from non-fiction books and online materials, featuring audio lectures generation, AI chat, and a real-time voice assistant.
Stack: AI Agents, Vector Search, LangChain/LangGraph, OpenAI, STT/TTS, TypeScript, Python, React, React Native, Node.js, MongoDB, Redis, Kubernetes, GCP.
- Composer AI
Built a Gmail-integrated AI agent that reads thread context, retrieves relevant information (past emails, sender history, calendar events), and generates tone-aligned replies using LangGraph + OpenAI.
Stack: AI Agents, LangChain/LangGraph, OpenAI, Browser extensions, TypeScript, Node.js, NestJS, React, MongoDB, GCP.
Co-founder & Product Engineer (CTO), Launcher Labs (YC W21, formerly Newzip)Jan 2021 – 2024
- Built and launched 7+ products across Web3 and consumer apps, including Launchcaster, Frameboard, and Newzip Livestream; owned CI/CD, Kubernetes deployments, and hiring.
- Architected a reusable backend framework (Node.js, GraphQL, MongoDB, Redis, RabbitMQ) that let the team ship features daily and pivot quickly through MVP phases.
Front-end Tech Lead, MokriyaAug 2018 – 2021
- Architected and led the build of the kw.com property-search front-end from scratch, hands-on alongside a team of 5 engineers, delivering a performant, scalable UI in React and GraphQL for one of the largest real estate platforms in the US.
Founder & Product Engineer, Liberbee (independent venture)Aug 2017 – 2018
- Founded and built a marketplace to buy, sell, and swap books; built it end-to-end — REST API (Node.js, Express, MongoDB, Elasticsearch) and an SSR React/Redux web client.
Full‑stack Developer & Product Researcher, Forbytes2012 – 2017
- Led front-end architecture for Trademax, one of the Nordics' largest furniture retailers, growing it into a multichannel e-commerce platform with custom back-office tools.
Education
Lviv Polytechnic National University2006 – 2010
- Master's degree — Computer Science