{
  "id": 18283674,
  "title": "Most Smart Devices Should Have Two AI Brains, One That Senses And One That Thinks: Nikunj Kotecha, Senior Machine Learning Engineer, Coastaloupe",
  "url": "https://urgent.news/2026/11/05/most-smart-devices-should-have-two-ai-brains-one-that-senses-and-one",
  "topic": "ai",
  "section": "AI",
  "published": "2026-11-05T09:21:17.000Z",
  "source": {
    "name": "Free Press Journal",
    "slug": "free-press-journal",
    "url": "https://www.freepressjournal.in/focused-view/most-smart-devices-should-have-two-ai-brains-one-that-senses-and-one-that-thinks-nikunj-kotecha-senior-machine-learning-engineer-coastaloupe"
  },
  "original_language": "en",
  "account": "Nikunj Kotecha, a senior machine learning engineer at Coastaloupe, believes that most smart devices should have two AI brains: one for sensing and the other for thinking. He explains that most smart devices currently rely on large processors that remain awake all day, consuming battery power to confirm that nothing has happened. Kotecha proposes dividing AI work between processors to optimize their usage.\n\nKotecha's career began at the Rochester Institute of Technology (RIT), where he focused on using AI to translate sign language. Later, he worked on AI chips at BrainChip, where he demonstrated that good models need to fit the devices they run on. Today, he works at Coastaloupe, an AI engineering solutions company in California, focusing on making AI run well on small devices.\n\nIn an exclusive interview with The Free Press Journal, Kotecha discussed his article, \"Two AI Brains Are Better Than One,\" which advocates for this heterogeneous computing approach. He explained that this method is already in use in devices like smart doorbells, earbuds, and factory sensors. Most other devices use a separate, larger processor for AI tasks, while Akida, a low-power chip designed by BrainChip, could be integrated into these devices to save power.\n\nKotecha also mentioned that BrainChip released the Symphony Community Akida Bundle, free software that brings Akida chips into IBM Spectrum Symphony. This integration allows small AI jobs to run on Akida chips more efficiently, saving power and reducing the need for separate processors.\n\nFurthermore, Kotecha is involved in Handshake AI's MOVE program, which connects experts with AI labs to improve the reliability of AI models. He serves as a fellow in the program and a reviewer, helping to validate and improve AI models for better performance in real-world scenarios.\n\nIn a recent project, Kotecha worked on IndicMMLU-Pro, a benchmark that tests AI models in Indian languages. He emphasized the importance of measuring AI performance in multiple Indian languages, as most AI models are currently evaluated in English. The benchmark consists of over 12,000 questions in nine Indian languages, with more than 7,600 downloads and publication in a research paper by the Association for Computational Linguistics.",
  "summary": "Artificial intelligence is moving out of large data centres and into phones, earbuds, doorbells and factory sensors, many of them running on small batteries. In an exclusive conversation with The Free Press Journal, Nikunj Kotecha, a California-based AI and Machine Learning Engineer who works on brain-inspired AI chips, talks about dividing AI work between processors, making AI work in Indian…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}