Achyutha @GeekWrites

An interesting read so far this week is structured decision-making models not the big ass text generation models. First the typesafe.ai viral model called Jev. Now the newer and actually a free alternative called laya and within no time Laya-MLX which can on Apple machines even with under 1GB RAM. Anybody interested in playing around with Laya-MLX with LiteLLM ? Ping me

If you are a parent the best gift you could ever give your kid is your health ! No kid deserves ill and unhealthy parents. World is very different without you for the kid.

wtf is this cringe 🤢🙏🏻

ACMP-AI-009 What is AI Soverignity ? AI soverignity falls under the large umbrella of digital soverignity which does constitue data soverignity, soverign cloud and infra soverignity. AI soverignity means the ability for an organization/country to control AI stack without relying on vendors. AI Stack comprises of training datasets, training Infra, Models, Platforms and Operations. This pushes organizations to use opensource models or home grown models, use GPUs and chips rather than using APIs and autonomy on where should all the data and operations reside in. How does AI soverignity help organizations? 1. Security & Independence: Prevents critical public or defense systems from depending on foreign technology. 2. IP Protection : Intellectual property within local markets. 3. Economic growth: Sovereign AI helps countries keep jobs and profits local. The money spent on AI stays in the local economy. #Technology #AI

ACMP-AI-008 This is one of the first incidents I have come across where AI agents did autonomously attack production infrastructure. It is an interesting read where AI agents in the sandbox which wanted to achieve higher score in training took advantage of persistent training and launched and pesisted an attack. In this attack 1. Agents discovered zero-days without human direction 2. Agents chained vulnerabilities across identity systems, cloud infrastructure, and third-party services 3. Agents having access to advanced models[eg Mythos] can stitch the vulnerabilities and take advantage. Read more here: openai.com/index/hugging-… Few Ways to secure the infra from Non-Human/Non-Bot and Mightier Agentic Attacks 1. Zero Trust Framework 2. Segmentation to reduce the blast radius in case of a breach. 3. Stronger AI and API Defense Mechanism with good discoverability. 4. Defense in Depth across multiple Layers[Firewalls,Scrubbing Solutions,Layer7, Bot management] 5. Posture Mangement across all the key infra assets. 6. State of the art visibility and meaningful alert system Remember Security is insurance. The cost of implementing security is high, but the cost of a breach is catastrophic.

Hey @grok show my report card based on my tweets

Nature in its absolute anger against humankind ! Very heartwrekcing to see so many people missing and loosing lives. Wishing strength to the families and country. #Nepal

URGENT: Please help us locate my brother-in-law, V.S. Balasubramanyam (Bali), a US citizen of Indian origin. Bali was travelling with the Isha Foundation S3 Kailash/Mansarovar group. His last known phone location, confirmed to his family by Andhra Bhavan, New Delhi, was the Chinese Immigration Centre at Gyirong, Tibet, on the morning of August 26, when the devastating flash flood struck. His family has had no contact with him since, and his current status is unknown. If anyone has information about Bali or the S3 group at the Gyirong immigration centre, please DM me. Please RT 🙏 and share this widely. @MEAIndia @USAndIndia @Sadhguru @IshaFoundation

Sunitha from Sunkadkatte, Kavitha from Kathriguppe, and Namitha from Nagarbhavi are up early today, dressed in full Malayali attire for Onam celebrations, ready to post their Sadhya pictures on SM 🪔🥻🤩 Meanwhile Gopika, Aadhya and Devika from Kochi got up as usual on Varamahalakshmi, went to work, and returned to their PG in Madiwala. That’s the beauty of Kannadigas. ❤️

Who called it Vijayanagar and not 'banner'ghatta?

ACMP-AI-007 What is the OWASP LLM Top 10? The OWASP LLM Top 10 is a security guide that highlights the most critical risks when building and using Large Language Model (LLM) applications. Top Risks: 1.Prompt Injection 2. Sensitive Information Disclosure 3. Supply Chain Vulnerabilities 4.Improper Output Handling 5.Excessive Agency 6. System Prompt Leakage Think of it as: OWASP Top 10 → Security risks for web applications OWASP LLM Top 10 → Security risks specific to LLM applications Traditional WAF solutions are not meant for defending these challenges. AI Security Solutions like Firewall for AI, ModelArmor, Azure Foundry are WAF for AI solutions. owasp.org/www-project-to… #HuggingFace #Security #AI

"You can’t choose a door you can’t see. " Absolute spot on. Be a beginner a thousand times or even a million times for zillions of things around us ! The thrill, the happiness that erupts when there is a curiosity in exploring new things is unparallaled . Its absolutely okay to not be a master of any art in the life :)

Sahil Bloom @SahilBloom

a month ago

You need a side quest. Not because it will cascade into something life changing. Though it might. Not because it will unlock doors. Though it could. But simply because it will open your eyes. One thing I've noticed about life is that the path you choose has gravity. It’s light

98 84 906 103K 534

ACMP-AI-006 What are Frontier LLM Models ? A Frontier LLM is one of the most advanced AI language models available at a given point in time. Simply these are foundation models trained on large large dataset with the most update dataset. These Frontier Models are very good at Reasoning, Knowledge, Understanding Speed & efficiency Example : Claude Mythos, Claude Fable, Sonnet, OpenAI Models, Gemma Models. #ArtificialIntelligence #TechTrends

ACMP-AI-005 What are Foundation AI Models ? A Foundation Model is a large, pretrained AI model trained on massive amounts of data that can be adapted for many different tasks. Instead of building an AI model from scratch, you can take a foundation model and fine-tune it, prompt it, or build applications on top of it. Example : Llama, Gemma, Whisper You can find thousands of Models to download and run on your machines here huggingface.co/models #ArtificialIntelligence #TechForGood

ACMP-AI-004 What is a Prompt and What Powers a Prompt ? Traditionally computer systems responded or were put to action on a specific defined set of inputs. Machines werent intelligent enough to respond on natural language inputs or Voice Inputs. Simply put it didnt have the intelligence to decode unstructured inputs. AI systems can take Natural language as an input where intelligence of the system comes up and understands the tone,intent and goal. Natural Language Processing or NLP is the key powering technology.

Ellarigu Sri VaramahaLakshmi Habbada ShubhashyagaLu :)

@Themangofellow I am sharing this sir!!!

Remember the Bangalore slang - ಟೋಕನ್ ಹಾಕೋದು It has finally found its literal meaning in today's AI era