Kevin Chiang

Xian Ke, this is incredible! The speed at which you and Vishal Punjabi leveraged crewAI, Streamlit, and various LLM models to create Finbot, a powerful company analysis tool for investors, was mind-blowing. ICYMI, Finbot is here: https://lnkd.in/eSQD69qA I couldn't help but chuckle when I saw the Finbot name - it reminded me of the Wall Street Bets crowd. Maybe you can get them to yolo their life savings based on Finbot's analysis? (Kidding, of course. Or am I? 💎🙌) But in all seriousness, your enthusiasm and passion for AI and LLMs is contagious. It's engineers like you who are pushing the boundaries of what's possible with AI and paving the way for a brighter future.🏆 I'd love to extend an invitation to any other software engineers from the bootcamp to join our Devs Do AI Discord server. It's a great place for engineers without an ML / AI background to connect with like-minded devs, share knowledge, and collaborate on LLM projects. DM me for the invite link. 🤝

13

Talking about LMMs and Astrophysics soon! :)

7

Highly recommended book...

9

🌟** Hi friends! Need 2 Expert Competitive Competitors **🌟 I’m excited to share that I’ll be competing in the CALICO Spring '24 programming contest hosted by UC Berkeley on November 16, 2024! 🧑‍💻💡 This competition is all about solving tricky algorithm problems in a limited time. I’m looking for 2 skilled competitive programmers to join my team. If you enjoy solving complex problems and want to collaborate on this challenge, let’s team up! Here’s a quick overview of the competition: - There will be 6-12 problems of different difficulty levels, and we have 3 hours to solve them as a team. - You can code in C, C++, Java, or Python3. - The problems will test both your coding skills and how efficiently your code runs. - The team with the most points wins, and in case of a tie, the fastest correct solutions will decide the winner. If you’re experienced in competitive programming on platforms like Codeforces, LeetCode, or CodeChef, I’d love to work together! 🚀 This is a great chance to improve our skills, learn from each other, and have fun competing. Send me a message if you’re interested in joining the team. Let’s make this competition one to remember! 🔥 Read this pdf for more info: https://lnkd.in/dBVTe2xZ Competition website: https://lnkd.in/dnvqUCd6 #CompetitiveProgramming #CALICO #CodingCompetition #TeamBuilding #ProblemSolving

30

6 Comments

The Wall Street Journal just released its list of the 250 best-run companies in the U.S. Here’s what stood out to me: 1. Many of these industry leaders are not just shaping the future—they’re our partners in tackling some of the toughest challenges in cybersecurity. 2. As cybersecurity leaders know, staying at the top means protecting everything you’ve built. That’s why I’m proud of the role we play in helping these companies secure their operations, protect customer trust and stay resilient in the face of evolving threats. Whether you’re already leading the pack or on the path to getting there, the whole team here at Arkose Labs is to help. Let’s make security your strongest advantage. #leadership #BestRunCompanies #AI #GenerativeAI

1

This weekend, I dove into something I’ve been curious about for a while: RAG (Retrieval-Augmented Generation). With the availability of great LLMs for coding, I figured I could actually complete an end-to-end RAG implementation in a weekend—something that normally wouldn’t happen. And it turned out to be true! LLMs helped me generate boilerplate code and accelerate progress, but there were still a few unexpected challenges I had to work through along the way. One of the biggest bottlenecks I encountered was resource constraints—running everything locally from my laptop was slow, especially since I had to summarize a large number of Confluence pages without external LLMs to protect company data. This experience reinforced my belief that, for organizations to successfully leverage LLMs, developers need access to either company-hosted models or the hardware to run local models effectively. Without these, teams might struggle to fully unlock the potential of these technologies. Despite the challenges, I really enjoyed this project and learned a lot about embeddings, vector stores, and the importance of summarization for good RAG results. It seems to me that building a RAG is one of those foundational steps in working with LLMs. Next on my list? Trying out GraphRAG, because I’ve got a love (possibly irrational, but probably not) for graph networks—they just align so well with how I think about things. Looking forward to experimenting more with LLMs and finding new ways to build cool stuff! Also, the diagram from this medium article helped spark some inspiration, though I didn’t use the code from it, so I figure I’ll share it: https://lnkd.in/eJZQskm9

9

Last week marked my last week at Voltron Data, after a massive round of layoffs that impacted half of the company. Two and a half years ago, I joined Claypot AI (@Voltron Data) as the first employee. Working at an early-stage startup and building something from the ground up was everything that I imagined it to be, if not more. I worked across functions and my responsibilities gradually evolved from a machine learning engineer into backend and infrastructure. I worked on database and API designs, orchestration and control plane, benchmarking Flink workloads, and building a compilation layer to parse feature declarations into optimized Flink jobs. After Claypot AI was acquired by Voltron Data, I continued leading the development of streaming at Voltron Data and worked on bringing the first three streaming backends into the open-source project Ibis. In the two and a half years, I learned so much and met many brilliant minds and some of the kindest and most incredible people. There were hard days just like any other startup, and I would’ve never predicted where I was going to end up, but I couldn’t have been more grateful for the opportunity. It's time for new challenges and I am actively looking for a new position! I have mostly worked as a machine learning engineer or software engineer in backend/infrastructure/streaming, but am open to new roles and opportunities. Would greatly appreciate any leads! #opentowork #layoff #startup #softwareengineer #machinelearning

320

22 Comments

Fantastic news today on the general availability of Amazon Q. Amazon Q is the most capable generative AI-powered assistant for accelerating software development and leveraging companies’ internal data. It eliminates tedious work for developers and employees across organizations. #AmazonQ #GenerativeAI

15

Excited to share a preview of the AI Prescreening Assistant I’ve been developing! This tool prescreens candidates via calls and has incredible potential in Customer Support, Sales, and Marketing. Current Architecture: Twilio for calls Speech-to-Text models for speech to text conversion. Text-to-Speech models for text to speech conversion. LLM's for response generation. With OpenAI Releasing Real-time API, this entire process will shift to a seamless Speech-to-Speech model, revolutionizing real-time conversations! DM if you want be beta Tester of Prescreening Assistant. :) #AI #Twilio #VoiceAI #OpenAI #Innovation #Prescreening

61

9 Comments

📈 Retrieval Augmented Generation (RAG) is increasingly becoming recognized as a critical technology for enterprises exploring advanced Generative AI solutions. Madison Lee and I have written an article that provides an overview of various RAG implementation strategies. Whether your organization is seeking - - No-code solutions for streamlined and accessible implementation, - Low-code options offering a blend of flexibility and ease of use, or - Full-code approaches for ultimate control and customization, Our guide demonstrates how IBM watsonx.ai can support all of these methods. Check out our article to discover the most suitable implementation method for your enterprise. Link: https://lnkd.in/g4Af3JEQ

22

1 Comment

🌟 Building Retrieval-Augmented Generation (RAG) Systems with LLMs and LangChain 🌟 I've been diving into how Large Language Models (LLMs) like GPT, Cohere, and Ollama, along with vector databases such as Pinecone and Qdrant, can create highly accurate and context-aware AI systems using LangChain. This innovative approach is known as Retrieval-Augmented Generation (RAG). 🔑 What is RAG? RAG is an AI architecture that enhances text generation by integrating information retrieval. It retrieves relevant data from external sources through a vector database, enriching the text generation process to ensure more informed and precise responses from LLMs. ⚙️ Steps to Build a RAG System: Create Embeddings: Use LLMs like GPT or Cohere to generate vector embeddings from unstructured data (e.g., text). Store Embeddings: Save these embeddings in a vector database like Pinecone or Qdrant for fast, similarity-based retrieval. Query Processing: Convert user queries into embeddings and retrieve similar documents from the vector database. Augment Generation: Feed the retrieved documents into an LLM via LangChain, enhancing the LLM's response generation with real-time data. LangChain Integration: Use LangChain to connect LLMs with vector databases, creating seamless workflows for retrieving and generating context-aware text. 🔍 Why Use RAG? RAG significantly boosts accuracy and contextual understanding by grounding LLM outputs in external, up-to-date information. This approach reduces hallucinations—instances where models generate incorrect facts—and provides contextually relevant responses. It’s particularly useful for applications like knowledge assistants, content generation, and recommendation systems. 🧠 Technologies Involved: LLMs: GPT, Cohere , Ollama are used for embedding generation and response generation. Vector Databases: Pinecone, Qdrant, and ChromaDB are utilized for storing and retrieving embeddings based on semantic similarity. LangChain: This framework connects LLMs with vector databases, enabling smooth RAG pipelines for enhanced AI responses. 🚀 Applications: Knowledge Retrieval: Systems that pull context-aware, real-time data for addressing complex queries. Personalized Content Generation: Generate accurate and fact-based content by retrieving the latest information. AI Assistants: Use retrieval-augmented LLMs to build intelligent assistants that provide relevant and timely information to users. Combining LLMs with vector databases and LangChain enables the creation of intelligent, scalable, and contextually aware systems for AI-driven applications. Exploring these emerging technologies has been an exciting part of my internship journey! Resources to dive deeper: LangChain Docs: https://docs.langchain.com Pinecone + LangChain Integration: https://lnkd.in/dZNbEZD9 OpenAI Embeddings Guide: https://lnkd.in/dxzCfzfV #RAG #LLMs #LangChain #VectorDatabases #AI #KnowledgeRetrieval #GPT #Cohere #Ollama #Qdrant

9

This article highlights some of the important and mirthful work I had the good fortune to take part in while working at Groundlight.ai. So happy to see it widely recognized.

5

Exciting news for anyone interested in building their own Large Language Model (LLM) from the ground up! 📘 If you're ready to dive into the process and deepen your understanding of the technologies behind LLMs, I highly recommend checking out this book: "Build Your Own Large Language Model from Scratch". This book is perfect for those who already have a foundation in programming, especially in Python. It guides you step-by-step through the intricate concepts, from preprocessing to model training, that are essential in creating an LLM. For me, it was both an enlightening read and a hands-on experience that boosted my understanding of this cutting-edge field. If you're ready to take your skills to the next level and create your own LLM, don’t miss out on this resource!

17

3 Comments

I heard this “critic” from serveral experts in the field and it resonates with my personal experiences and from my observations of ppl. We grossly overestimate and zone in on LLMs / gen-“ai” https://lnkd.in/e242rQCT

3

Ever heard of the California Gold Rush? It's one of the most significant events in American history, beginning in 1848 and lasting until 1855. It all started when James W. Marshall discovered gold at Sutter's Mill in Coloma. News of the discovery spread quickly, and soon thousands of people, known as "forty-niners," rushed to California from the U.S. and abroad. The rapid influx of gold revitalized the American economy, and the swift population growth enabled California to achieve statehood quickly as part of the Compromise of 1850. But here’s the real question: Did all the gold seekers strike it rich? Not exactly. While a few found success, many did not. If we break down this historic event, three key elements were involved: Product, Process/Platform, and People. ⛏ Product is the gold. ⛏ Process/Platform refers to the services and infrastructure that supported the search for gold—shops selling mining tools, restaurants, and hotels catering to the "forty-niners." ⛏ People are everyone who understood the value of gold. Among these, people were the most crucial for any business to thrive. Without them, there would be no demand for gold. Interestingly, while many gold seekers didn't strike it rich, those who provided tools and services—the middlemen (Process/Platform)—were the ones who profited most, without ever digging for gold themselves. Now, fast forward to today, and we’re witnessing a similar phenomenon, but this time it’s the AI Rush. 🤖 The product is AI. 🤖 The platforms are the companies like NVIDIA, AMD, and others that offer the tools and technology to harness AI’s power. 🤖 And, once again, the people—those who understand AI’s value—are the driving force behind this new wave of innovation. Just like the Gold Rush, the success stories won’t just come from those who are "mining" the AI itself but from those who provide the tools and platforms to make it happen. #AI #Innovation #TechRevolution #BusinessTransformation #AIRevolution

24

🚀 Introducing SpectrumX: Next-Generation Multi-Device Data Acquisition Software 🚀 After 8 months of development, I’m excited to share SpectrumX (Alpha), a cross-platform, high-performance, and fully reactive solution built from the ground up with TypeScript, React, and Rust. Key features: ⚙️ Device-API: Developers can write their own device plugins in TypeScript and seamlessly integrate them into the Spectrum runtime. 📊 WebGL charting: Capable of rendering over 10 million points per second on average hardware, maintaining smooth performance at 40 FPS. 🧵 Multi-threaded architecture: Handles data acquisition, database operations, and real-time calculations with optimal performance and stability. 🖥️ Cross-platform support: Runs on Windows, Mac, and Linux, ensuring wide accessibility for different environments. 🎨 Custom accessible UI components: Built with React for a responsive and user-friendly experience. With SpectrumX, we are trying to push the boundaries of what real-time data acquisition and analysis software can achieve. Check out the promo video to see it in action! #softwaredevelopment #typescript #rust #reactjs #webgl #dataacquisition #crossplatform

17

1 Comment

🌟 Exciting Milestone 🌟 I'm thrilled to share that I've successfully completed the Building Generative AI Applications with Gradio course! This accomplishment has significantly boosted my skills in developing user-friendly AI interfaces utilizing Gradio. The journey of learning, from creating AI models to designing engaging applications, has been truly rewarding and enlightening. #GenerativeAI #Gradio #AIApplications #LearningJourney #AIInnovation Check out my achievement: https://lnkd.in/g-FHMCe8

4

📢 Wait — Folders Aren’t Really Folders in Amazon S3 Buckets? Most of us think Amazon S3 works like a traditional file system with folders and directories. But here’s a surprise: those "folders" are just visual aids! 😲 Amazon S3 is a cloud storage service, while an S3 bucket is a container for storing objects in S3. It’s a flat storage system, where each object is given a unique key. The forward slash (/) is simply a delimiter that creates the appearance of a folder structure. In reality, S3 has no hierarchical folders—just unique object keys. ✨ This design enables faster access, scalability, and flexibility, helping developers organize and retrieve data seamlessly. Want to dive deeper? Check out the full article here 👉 https://lnkd.in/ddGHsfwK Did you already know this about S3 buckets? Let me know! 👇

6