AI/ML, Quantum, Security | Author | Innovator | Advisor | MSc. Observational Astrophysics | MSc. Mathematics | CEng | CSci | MBCS-CITP | Google Cloud Architect | Volunteer Cricket Coach
Grateful to be invited to the Estonian Embassy to attend an event on public and private sector engagement for the Technology Partnership Cooperation Agreement, signed in Dec. 2022 between Estonia and United Kingdom. Estonia is a highly digitalised country with deep focus on further improvements via visionary policies and tech. Our very gracious host, the Estonian Ambassador to the UK, H. E. Viljar Lubi was hospitality personified and I am so grateful for the time, patience and discussions with His Eminence. Speaking to the Ambassador - who very kindly ensured time for all attendees, the panels moderated by Jeegar Kakkad of Tony Blair Institute, listening to the policies on AI for governance from Ryan Shea & team - was a unique experience. One of the key focus was on how to make LLMs more efficient and relevant. The other area in a digiltalised world, perhaps more important than anything else, is cybersecurity. Security can affect AI and other domains severely, as the complex threat surface of the AI ecosystem is exposed to attacks such as poison data and prompt injection attacks, besides the looming future threats from faster computing systems such as quantum computers. At SandboxAQ, we have been using RAG based LQMs (Large Quantitative Modles). Retrieval Augmented Generation (RAG) with Large Quantitative Models (LQMs) is a step forward addressing some of the limitations of LLMs such as hallucinations and out-of-date training data. Many LLM vendors don’t retrain their models frequently enough due to the time and cost involved. As a result, the data the LLMs draw on, is commonly not up to date. And, LLMs have no access to private data within organizations – the very data that enables the personalized and targeted responses. RAG is a GenAI framework that improves the response capabilities of an LLM in a cost-effective manner by retrieving and injecting into the LLM, up-to-dated, trusted data from customized, private knowledge bases. SandboxAQ is using their transformative impact in fields such as healthcare (for example, AI simulation in drug discovery), automotive, defense, and more. In an interview on Bloomberg TV's Wall Street Week hosted by David Westin, SandboxAQ CEO Jack Hidary discusses how : https://bit.ly/4co1Tyw LQMs: https://lnkd.in/eAurED3W AQNav: Magnetic Navigation that has been selected as one of the best inventions of 2024 by Time Magazine: https://lnkd.in/eS8QVaSz Cybersecurity: Crypto-agility and why it's so important from SandboxAQ's Cryptography Cafe: https://lnkd.in/ePjEsnSM AQMed: https://lnkd.in/e-Zz-AzH And, for anyone interested in getting their hands dirty with RAG, LLM, CUDA and GPUs, NVIDIA has a wonderful, hands-on training that I am going through now myself: Building RAG Agents with LLMs (need some Python background): https://lnkd.in/eF-WTuH7
Data Scientist at Tata Consultancy Services (TCS) | MTech, Data Science & Engineering, Birla Institute of Technology and Science, Pilani
2wThanks for sharing, very informative and helpful