Nirmala Sistla

Seeing a lot of excitement for row-level security (which is amazing, no doubt!) but my team is more excited about column-level security. Being in banking, a lot of records are peppered with PII and the governance over who gets to see what (combination of) columns can get messy. With CLS, we can implement fact and one-big tables without having to resort to view gymnastics (i.e. using views to hash columns depending on the viewer) or maintaining multiple pipelines for the same table with different audiences. Just another win for Unity Catalog and Databricks.

4

Unlocking the Power of LLMs: a framework for adoption Last week at Ai4 - Artificial Intelligence Conferences I participated in the Making Use of LLMs panel, moderated by AI legend Manoj Saxena, with industry leaders Christopher Stephens, Dan Mackey  and Raj Aikat. We discussed various strategies effectively leveraging LLMs in practical applications. My contribution focused on the progressive adoption model Scotiabank is using to walk/crawl/run with LLMs. Each successive stage in this model is more complex to implement than the previous one, but also yields greater benefits. The stages are: 1. Public Chatbots: While limited in their capabilities and requiring controlled use, platforms like Microsoft Copilot and ChatGPT can be valuable tools for numerous applications. 2. General Productivity Tools: Tools like Copilot 365 (and others) access your documents, emails, and chats, transforming how we work. They can draft emails, generate code, and summarize complex information, significantly boosting productivity. 3. Custom Tools for Specific Teams: this is where it gets exciting! Scotiabank is developing LLMs tailored to the needs of specific employee groups, saving time and boosting efficiency. 4. Back-Office Automation: LLMs are also revolutionizing back-office processes. For example, imagine a chatbot summarizing customer interactions before transferring the chat to a live advisor – this streamlines workflows and improves customer experience. At Scotiabank, we believe LLMs have the potential to transform every aspect of our operations. By strategically implementing these powerful tools, we're driving innovation, enhancing productivity, and creating a more efficient and customer-centric experience. Photo credit: César Francisco Landa Paredes

251

12 Comments

Neo4j's recent blog post on 𝐊𝐧𝐨𝐰𝐥𝐞𝐝𝐠𝐞 𝐆𝐫𝐚𝐩𝐡-𝐆𝐞𝐧𝐀𝐈 𝐢𝐧𝐭𝐞𝐠𝐫𝐚𝐭𝐢𝐨𝐧 highlights the potential of 𝐆𝐫𝐚𝐩𝐡𝐑𝐀𝐆 to enhance AI applications' accuracy, contextual understanding, and explainability. While 𝐆𝐫𝐚𝐩𝐡𝐑𝐀𝐆 demonstrates superiority over vector-only RAG in these aspects, its implementation efficacy is use-case dependent. Large-scale data scenarios benefit from graph integration, but extensive graph traversal may introduce latency, potentially offsetting performance gains. Optimal implementation necessitates strategic query routing between graph and vector databases to maximize system efficiency. #KnowledgeGraph #GenAI #RAG

34

2 Comments

Good article on practical tips to improve the #RAG pipeline - across indexing, retrieval and generation. • Indexing - level up with simple, semantic or #LLM based chunking • Retrieval - implement hybrid search, combining vector and keyword searches • Query Rewriting - optimize queries for better retrieval • Metadata - boost retrieval accuracy with rich contextual data • Autocut - filter out irrelevant information from retrieved content • Reranking - prioritize the most relevant content, enhancing retrieval precision https://lnkd.in/eJMvu3XJ

7

Organizations have insufficient technical capabilities for the speed and scale of their ML model implementations. My colleague Sumit Agarwal just published a very helpful research with a capability view and sample vendors for defining a modular, flexible and scalable MLOps solution. https://lnkd.in/gnB_T9bq #MLOps #LLMOps #AI

149

17 Comments

A recent #HBR article gives real-world examples showcasing how #GenAI is leveling the playing field. It's interesting to see the various ways where this transformative tech is being used. In this context, I often hear how GenAI is enabling skill democratization. It though doesn't level the playing field by enabling anyone to perform tasks across various domains without background or training. Instead, as the examples in the article point out, it augments existing skills, making it easier to accomplish tasks by lowering the barrier. HBR article - https://lnkd.in/eiSQVJBv

11

1 Comment

Looking forward to Amazon Research Days on Nov 21 (Thur) featuring an exciting lineup of talks from Industry (Anthropic, Sarvam AI, Amazon) and Academia (UNC Chappel Hill, Australian National University). Not surprisingly, this year, we have multiple talks on training foundational GenAI models, and applying multimodal GenAI models to a wide range of domains such as product image generation, 3D construction, speech recognition, etc. Amazon Research Days is an ML conference held annually that brings together researchers and practitioners from academia and industry to exchange and brainstorm new ideas, learn about state-of-the-art ML techniques, and network to build a strong ML community. #amazon #amazonscience #amazonresearchdays #ai #ml

427

6 Comments

We all remember that 'aha' moment with ChatGPT. Are you ready for the Agentic AI moment? Watch this quick 5-minute video to dive into Agentic AI, explained by none other than Andrew Ng

42

A great read for anyone building generative AI applications. It touches on actual pain points and the excellent advantages that GenAI applications have. Some highlights include 1. How LinkedIn determined its use case 2. Actual components of the application ( RAG and agents) 3.Automatic evaluation.(Personally this is a favorite of mine) 4.Technical evaluation of latency and throughput . Please read! It’s fascinating. Also found this via Chip Huyen ‘s wonderful post.

4

Value capture has always fluctuated between applications and infrastructure. What will we see getting commiditized first in the ML world - compute infra or applications? Read my thoughts on this here : https://lnkd.in/gD3N7dHX #strategy #hyperscale #scaleups #bizgrowth #foundersadvice #bizmodel #innovation #fintech #insurtech #tokenization #productmanagement #web3  #TMinsights

5

GenAI leads to new economies of scale by generating artifacts in bulk or by helping humans to generate them faster. There is plenty of room for commercialized solutions to create new and original value with those #GenAI artifacts. 

96

3 Comments

How are people experimenting with #genAI? A recent study published in #HBR highlights the usages beyond corporate settings. Top 5 includes idea generation, search, and providing editorial insights. Notably, the list also includes usage for law, medicine and coding. While the appeal of using genAI to these expensive services is evident, not sure how well the associated risks are understood. https://lnkd.in/eMD5dP9k #AI

13

Smaller, more powerful LLMs on edge devices, combined with advancements in hardware, are paving the way for the next generation of robotics. In my opinion, this is the way forward. By enabling versatile, general-purpose robots that can operate efficiently without relying on cloud connectivity, we’re unlocking a smarter, more autonomous future. In the latest episode of AI Success Factors, I shared my thoughts on how AI and robotics are redefining industries like healthcare, logistics, and manufacturing. I also provided practical tips on leveraging these advancements to stay ahead in this rapidly evolving space. If you’re curious about: > Why edge computing is a game-changer for robotics > How AI is driving innovation in IoT and hardware > Real-world examples of robotics reshaping industries …then you won’t want to miss this episode. Listen to the full podcast here: Apple Podcasts: https://lnkd.in/g_y35bqc YouTube Music: https://lnkd.in/gK2b8B6W Spotify Podcasts: https://lnkd.in/gbgpeXYQ #ai #robotics

17

1 Comment

AI Capabilities Plateauing? Is "Bigger is Better" Losing Its Edge? A recent article by Reuters highlights these challenges and the industry's pivot toward new paradigms, such as reasoning and inference-time computation. The focus is shifting from brute force scaling to models that innovate and think independently. For years, scaling up pre-training—feeding vast amounts of data into ever-larger models—has been the cornerstone of AI advancements. But according to Ilya Sutskever, co-founder of OpenAI and Safe Superintelligence (SSI), we may be hitting a plateau. Recent findings suggest that simply increasing model size is yielding diminishing returns. Yann LeCun puts it succinctly: "Unless the learning system can 'invent' solutions and answers to new inputs by itself (through an ability to reason) instead of merely receiving and adapting previously-learned answers, the plateau will be with us." Implications for the Industry - Are we ready for AI systems that think and reason rather than memorize and adapt? - How do we redesign learning paradigms to enable true reasoning and problem-solving? This isn’t just a technical shift—it’s a philosophical one. AI’s evolution depends on its ability to move from scaling to reasoning. Reuters Article: https://lnkd.in/e2uyZVVW #AI #MachineLearning #ScalingLaws #Innovation

8

3 Comments

I remember when this all started. Faris - invited me to catch up at our favorite spot around Blok M. I had no idea he’d just wrapped up a procedure at Pondok Indah Hospital! (He really knows how to add drama to our catch-ups.) We’d been on separate paths and projects since "the winding down project", but there he was, reminding me of our dream to ride this AI wave together again. It was a plan we’d hatched with Sahal Zain & Gurnoor Dhillon when the world of AI was still mostly theoretical and jargon-heavy. Somehow, the conversation continued with Gurnoor Dhillon while he was waiting for his meeting with someone who would eventually become one of CalvinBall’s key early adopters. It felt like everything was coming together, like a series of seemingly random dots connecting to form a bigger picture. Seeing how far we’ve come as a team at CalvinBall Technologies is truly exciting. What started as theoretical discussions has now evolved into practical Gen AI solutions that are solving everyday problems and hold endless possibilities to shape the industry. 👨‍🚀 👩‍🚀 Here’s to pushing boundaries and achieving new milestones together! Gurnoor Dhillon, Sandeep Ramesh, Faris -, Sahal Zain, Eva Zuliya, Ratu Nadiah Khairunnisa 🚀 #GenAI #AIInnovation #TechJourney #ArtificialIntelligence #TechTransformation #CalvinballTech #Calvinballers

55

2 Comments

As AI workloads move from POC to Production, understanding the cost implications for running that workload is imperative for success. The AI working group in the FinOps Foundation put together this guide to help you be successful. I had a great time working on this with the team and I’m very proud of our results and proposals. Brent Eubanks ☁️ Josh Collier Ermanno Attardo Alireza Abdoli Max Audet What have you been doing to manage your AI costs? https://lnkd.in/gApubsYa #ai #costmanagement #finops #cost #accuracy #performance

47

5 Comments

Pondering on the possibilities of learning and education as AI and emergent technologies continue to evolve. Just tried Illuminate and summarized Stanford's CS224n (NLP with Deep Learning) week 1 course work (seminal papers on word representations) into a 6 minute audio clip! If this is the beginning (as a lot of experts in AI are saying), the future seems bright 🙂 https://lnkd.in/gqpS7ysn < the link is limited to first 50 listeners >

12