AI Engineer focused on building production-grade AI systems in regulated, real-world environments.

Over the last 18 years, I’ve worked across startups, enterprises, and government, designing and operating systems where correctness, reliability, and usability matter. In recent years, my work has centered on LLM-based and multi-agent architectures, with a strong emphasis on evaluation, observability, and long-term ownership in production healthcare systems.

I gravitate toward hands-on Staff / Architect IC roles, where I can own complex systems end-to-end, define technical direction, and build shared AI platforms that other teams rely on — without politics or ivory-tower design.

• Multi-agent systems for voice and text-based AI applications
• Production AI platforms used by multiple engineering teams
• LLM-based architectures designed for reliability and debuggability
• Automated evaluation (simulation-based testing, precision/recall/F1)
• RAG-backed systems with measurable quality and controlled behavior
• Platform abstractions enabling model and vendor flexibility
• Product-aware engineering, aligning AI rigor with real user needs

• Evaluation is a first-class concern, not an afterthought
• Non-determinism requires explicit control and observability
• AI systems must be measurable, testable, and evolvable
• Shipping is not the end — post-launch reliability and evolution matter
• Architecture should reduce repeated decisions and operational risk

• Led architectural resets from single-prompt assistants to multi-agent voice AI platforms in production healthcare environments
• Built automated evaluation frameworks simulating hundreds of real user interactions to detect regressions early
• Designed HIPAA-compliant AI platforms enabling multiple teams to safely build and iterate on shared infrastructure
• Extensive experience with voice AI, RAG, LLMs, and AI observability
• Long-term ownership of systems beyond launch, including reliability, evolution, and developer enablement

• Reliable and evaluable AI systems
• Agentic architectures and orchestration patterns
• AI platform engineering
• Voice interfaces and conversational systems
• Bridging the gap between AI demos and real-world production

• LinkedIn: https://www.linkedin.com/in/ivanviragine
• GitHub: https://github.com/ivanviragine

This profile reflects hands-on work on real systems, operated in production.
If you’re looking for hype, benchmarks without context, or prompt-only demos, this is probably not a good fit.