Ivan Shvedov's Work | ContraWork by Ivan ShvedovIvan Shvedov
I build AI products, MVPs, web apps and automations
Built Nashday.ru, a SaaS platform that allows couples to create personalized wedding websites, send digital invitations, and manage guest RSVPs.
This is my independent product, which I designed and developed from the initial idea to a working production platform using AI-assisted development.
What I built:
- Defined the product concept, user journeys, architecture, and monetization model.
- Developed the frontend, backend, database, and user authentication system.
- Built a wedding website editor with customizable templates, photos, event schedules, locations, and guest questionnaires.
- Implemented guest invitation management and RSVP tracking through a personal dashboard.
- Integrated payment systems and a checkout flow, allowing users to create invitations for free and pay before publishing.
- Handled infrastructure, deployment, testing, and production releases.
- Developed a wedding vendor directory and started building an AI-assisted SEO automation pipeline for organic growth.
Result:
Launched a complete SaaS product with a working website builder, user accounts, payment integration, and a monetization flow.
I handled the entire process independently, from product strategy and UX design to full-stack development and deployment. Led the development of Bayun.ai, a B2B SaaS platform that uses AI to analyze customer conversations, identify lost sales opportunities, and improve service quality.
What I did:
- Conducted 30+ customer interviews, identified business needs, and defined the product concept and MVP.
- Designed user journeys, product architecture, and the AI processing pipeline.
- Independently built the initial prototype and MVP to validate the concept.
- Organized pilot projects, collected and labeled speech datasets, and led the fine-tuning of our own speech recognition model.
- Benchmarked our model against Deepgram. After achieving better recognition quality, we switched to our own STT model, reducing processing costs.
- Integrated Mango Office calls and implemented customizable AI conversation scoring based on each client's internal standards.
Results:
- 30+ customer interviews
- 5 pilot projects
- 3 pricing discussions
- 2 commercial deployments
- Reduced speech processing costs through an in-house STT model
Took the product from initial idea and customer discovery to a working AI platform with paying clients. Built and launched a fully functional 3D kitchen planner for AMS Software in just 9 working days.
The challenge: Replace a third-party planning tool with an in-house solution and make the customer journey measurable and controllable.
What I delivered:
- Designed the product architecture and user journey.
- Developed the frontend, backend and PostgreSQL database using AI-assisted development.
- Built an interactive kitchen planner with 2D/3D visualization, configurable cabinets, room dimensions and project saving.
- Implemented onboarding, event tracking and conversion analytics.
- Handled testing and production deployment.
Results:
- MVP launched in 9 working days.
- Software download conversion increased from 6% to 10%.
- The company gained full control over the customer funnel and product analytics.
Took the project from requirements to a working production application, without a traditional development team. Designed and built EmTalk.org, an AI-powered Web2App product, independently from concept to production.
I was responsible for the entire product lifecycle, including UX/UI design, user journeys, frontend and backend development, database architecture, API integrations and deployment.
I also independently developed and integrated a Retrieval-Augmented Generation (RAG) solution to deliver context-aware AI functionality.
The project combines product design, full-stack development and AI engineering, without relying on a traditional development team.