Armenia’s growing tech community took another step forward recently as Krisp Technologies, one of the country’s leading AI startups, hosted a three-day engineering hackathon focused on artificial intelligence and large language models. The event, held from September 12–14, brought together 33 highly skilled AI engineers — both from Armenia and Russia — who were challenged to build a next-generation chatbot using a proprietary knowledge base. The winning team took home a US$5,000 (approximately 1.9 million AMD) grand prize.
According to Krisp, the event attracted over 80 applicants, with only 33 selected based on their AI expertise and ability to implement advanced techniques like Retrieval-Augmented Generation (RAG), GraphRAG, and multi-hop retrieval — methods typically used in frontier AI systems. Participants were divided into nine teams and given one mission: to build an AI-powered chat application that can quickly provide users with relevant information from a large and complex internal data system. The challenge required not only deep technical expertise in AI but also innovative problem-solving and system architecture skills.
The projects were judged by a panel of Krisp’s top AI engineers and Erik Davtyan, who served as the hackathon’s only external judge. Erik Davtyan is an AI engineer and GenAI specialist who was invited to judge for his recognized expertise in semantic search systems and AI-assisted tools. Most recently, he worked remotely with A.Team, a $55 million–backed New York startup, where he improved the company’s AI-driven search relevance by 60% using semantic reranking techniques, and increased the data pipeline reliability to 99.9%. He is also the co-creator of the ChatGPT extension for Raycast, a macOS productivity tool used by over 200,000 people.
The judges scored the projects based on system design, innovation, sophistication in applying cutting-edge AI techniques, scalability and efficiency optimization. The winning team was EXPM, followed by Ratatouille in second place, and HArm in third.
According to Davtyan, the event highlighted the strength of Armenia’s AI engineering talent. “The level of technical sophistication was high. The problem was intentionally open-ended, and it was interesting to see the creative and varied approaches that the teams took. All teams were able to build working prototypes in two days — the kind of agility you’d expect to see in teams at top U.S. or European companies,” he said.
He was particularly impressed by the runner-up team, Ratatouille, composed of researchers and ex-researchers from Nvidia. “In only 48 hours, they built an improved version of the already cutting-edge RAPTOR (Recursive Abstractive Processing for Tree-Organized Retrieval) algorithm. It’s impressive for algorithms of that complexity to be designed and executed in such a short timeframe.”
Davtyan also praised team HArm, saying, “While their algorithm wasn’t new, they mainly focused on the scalability of their solution and were able to reliably analyze very large files up to 100 GB in size.”
However, it was the winning team, EXPM, composed of engineers from leading tech companies like ServiceTitan, Portmind, and Develandoo, that impressed the judges the most. “EXPM’s algorithmic design and system architecture were production-ready in just two days,” Davtyan explained. “EXPM did this by layering AI multi-hop reasoning with intelligent query planning. They combined AI retrieval with classical data storage systems and came up with a reliable and scalable solution, which worked across a variety of use cases.”
Beyond the competition, the hackathon offered a clear snapshot of Armenia’s rapidly advancing AI talent, demonstrating just how capable these engineers are. With global companies like ServiceTitan and Nvidia, alongside national startups such as Krisp and Develandoo employing Armenian talent, the country is steadily shaping the future of AI-driven solutions. Armenia is emerging as a serious player in the Caucasus region’s technology landscape, and with continued investment in AI and advanced technologies, the nation’s startup ecosystem will be worth watching closely.
About Krisp Technologies
Founded in 2017, Krisp Technologies is an Armenian AI startup best known for its breakthrough noise-cancellation technology, which uses deep learning to remove background noise from online calls in real time. Krisp’s technology is used by millions of professionals worldwide and integrated into major communication platforms. The company continues to expand its focus into broader AI applications, including productivity tools and intelligent communication systems.
Lilith Avagyan
Follow NEWS.am Tech on Facebook and Twitter