Sai Harshavardhan

Research on agentic memory architectures at the Chair of Cognitive Systems; building data pipelines that reduce repeated records and irrelevant context in long-term agent memory.

Build research-data and experiment pipelines around Cadence Virtuoso and agentic automation for internal engineering workflows at the Institute for Information and Communication Technology (IIKT).

Only technical person in a small training company moving online. Launched an n8n workflow with a language model agent that chats with leads on WhatsApp, collects their details and hands them to a human; it went live with real prospects. Designed a verification step so staff sort leads as genuine, bot or low-intent. Built the landing page.

Built an air traffic control speech transcription pipeline with Whisper and DeepSeek post-processing at the Mission and Combat Systems Research and Design Centre (MCSRDC).