Oleksandr Zalizniak
Product Engineer & Backend Architect. This is my corner of the web: a bit about what I’m building, plus Applied AI Teardowns, where I reverse-engineer how applied-AI startups build their products from public signals alone.
Résumé — PDF · DOCX · LinkedIn · GitHub
About Me
Section titled “About Me”I’m Oleksandr. Right now I’m building Signus.ai, an AI-native contract intelligence platform powered by Sigi, an AI legal agent for generating, signing, analyzing, and tracking agreements at scale.
Before Signus, I spent 15+ years as a backend architect and full-stack engineer — designing cloud infrastructure and microservices on AWS across SaaS, e-commerce, finance, and telecom, most recently leading backend architecture at UserWay (~10k req/s across 15+ services).
I’m based in Geneva, NY, where I hike, rollerblade, and watch too much Kill Tony.
What I Do
Section titled “What I Do”- AI Product Engineering: Building AI-native products — LLM agents and contract intelligence.
- Backend Architecture: Scalable, resilient services in Java, Node.js, Spring, and Nest.js.
- Cloud Infrastructure: AWS at scale — ECS, Lambda, RDS, Redis — and microservices under heavy traffic.
- DevOps & CI/CD: Deployment automation and pipelines that ship safely and often.
- Technical Leadership: Leading cross-functional teams from idea to production.
Applied AI Teardowns
Section titled “Applied AI Teardowns”I publish Applied AI Teardowns — a library reconstructing how applied-AI startups build their products, from public signals only (job postings, blogs, docs, product behavior). The unit of value is dense, evidenced inference. Start with the Pallet teardown.
SpecCraft open source
Section titled “SpecCraft ”I maintain and contribute to SpecCraft — a semantic concept-mapping tool for learning and research. You build a graph of concepts and relations; AI agents (over MCP, or in-app chat) do the routine work — discovery, expanding ideas, finding associations, explaining — so you can focus on actually understanding a domain.
The idea: not a design canvas, not a notes app, but a thinking tool where the agent is a first-class collaborator on a shared semantic graph. Built with NestJS, Next.js, Postgres, and a remote MCP server.
Temporal & Durable Execution
Section titled “Temporal & Durable Execution”I’m a firm believer in Temporal. After ~1 year building on it, I’m convinced workflows-as-code is the right default for anything that has to survive crashes, retries, and long waits — it turns brittle distributed flows into something you can actually reason about.
Building Temporal-related tooling is my current priority — alongside going deep on Go, the language at the heart of Temporal’s ecosystem.
Everything Around the Model
Section titled “Everything Around the Model”Models get the headlines, but the work that lasts is everything around them — the layer that turns a clever demo into something reliable at scale. That’s where I’m going deep next, and I want to be genuinely expert at it.
What pulls me in: retrieval that keeps answers grounded instead of hallucinated, evaluation harnesses that prove a prompt or fine-tune actually helped, RLHF and feedback loops that turn real usage into training signal, and agentic systems that plan, call tools, and recover from their own mistakes.
Let’s Connect
Section titled “Let’s Connect”Whether you’re looking for a technical lead for your next project, someone to brainstorm innovative solutions, or just want to chat about the latest in tech, I’d love to connect.
Email: alex@zalizniak.com · LinkedIn · GitHub