I'm the founder of Agyn, the open-source management layer for AI agents. Agyn helps companies safely deploy the agents they've built — define an agent once, ship it across the business with isolation, secret management, spend controls, and access governance built in. Alongside the platform, I led research on multi-agent collaboration for software engineering and published the results: our system achieved 72.2% on SWE-bench Verified — the #1 result among GPT-5-based systems at the time of release, and evidence that coordinated multi-agent setups can outperform single-agent ones on real engineering work. I've spent the last eight years turning AI research into production systems at global scale. As an R&D group manager at Kaspersky, I co-invented 5 US patents in ML-based cybersecuri...