Data Science Dojo Karachi

Welcome to the community series with Weaviate! Vector embeddings are a cornerstone of modern artificial intelligence, transforming complex, abstract data into numerical representations that machines can process and understand. In this three-part webinar series, we will dive deep into vector embeddings, vector search, and the challenges of deploying ML/AI applications at scale.

Part 1: Introduction to Vector Embeddings
What we will cover:
- Why the 2017 Word2Vec paper revolutionized AI with embeddings.
- An overview of how modern embedding models work and their use cases.
- Practical demos on creating vector embeddings in Weaviate, OpenAI, and Huggingface.
- Key considerations for selecting the right embedding model, including model size and industry relevance.

Part 2: Introduction to Vector Search in Vector Embeddings
What we will cover:
- Understanding how mathematical similarity is calculated in vector search.
- Limitations of traditional search methods and how vector databases address them.
- In-depth look at ANN algorithms like HNSW and their role in optimization.
- Practical demo on implementing vector search with Weaviate over the Wikipedia dataset.

Part 3: Challenges of Industry ML/AI Applications at Scale with Vector Embeddings
What we will cover:
- Recap of vector embeddings and similarity from previous parts.
- How AI powers systems like internal search, recommendation engines, and RAG.
- Practical demos of real-world ML/AI applications using Weaviate.
- Emerging trends like AI agents and GFL and their impact on scaling AI.

Are you ready to unlock the full potential of LLMs by harnessing the power of agent-based collaboration? Join us for an insightful webinar as we explore how to build scalable, event-driven AI solutions using LlamaIndex. From theory to hands-on practice, this session will guide you through Agentic AI workflows that simplify complex problems and bring real-world AI applications to life.

Experience a live demo showcasing ReAct agents in action, seamlessly integrating reasoning and actions to interact with external tools, process data, and make intelligent real-time decisions. Gain practical insights on how to adapt these workflows to solve challenges in industries like finance, healthcare, and technology.

What You’ll Learn
- Discover how agent-based workflows can solve complex problems through collaboration.
- Build a multi-step workflow using LlamaIndex, from data intake to intelligent decision-making.
- Incorporate real-world data into your workflow to enhance AI responsiveness and intelligence.
- See how ReAct agents powered by Agentic AI, which combine reasoning and actions to interact with external tools, process data and make real-time decisions in the live demonstration.

Agentic applications are a fast-growing development paradigm where Large Language Models (LLMs) are utilized to do a lot more than summarize some documents. These applications can initiate actions (for e.g. save new releases to your Spotify playlist), trigger workflows, or be given full agency to determine how to achieve a certain objective on behalf of the user.

In this webinar, we will dive into the core concepts behind agentic AI applications, compare it to traditional SaaS systems, and examine how techniques — like function calling and planning – play a powerful role in building these applications.

We will also explore how you can leverage existing business functions and APIs to build powerful systems with Arch, the open-source, intelligent gateway server for intelligent apps. Through demonstrations and discussions, you will gain a clear understanding of how to approach agentic system architecture.

What we will cover

• Understanding tasks, workflows, and agency in agentic applications.
• Gain a high-level overview of when (and when not to build) agentic systems.
• Delving into intent detection, function calling, and planning to enhance task performance.
• Configuring functions and API definitions in Arch Gateway to build a real-world agentic application.
• A complete walkthrough demo of an agentic app retrieving real-time information in RAG scenarios, performing actions like submitting forms, calling APIs, and modifying application states (like handing off the conversation).