Let's Build a Startup - S2E4 - Build and Monetize your GenAI MVP with AWS and LanceDB

Let's Build a Startup - S2E4 - Build and Monetize your GenAI MVP with AWS and LanceDB

Join Chang She (LanceDB), João Galego, and Kevin Shaffer Morrison to discover tips and tricks to build genai applications without breaking the bank

Giuseppe Battista
Amazon Employee
Published Jul 1, 2024
Do you have a Generative AI startup idea but don't know where to start? Do you want to test that idea without breaking the bank? If so, this episode of "Let's Build a Startup!" was written for you!
With the help of Chang She–CEO and co-founder of LanceDB–and João Galego–Startup Architect at AWS–Giuseppe and Kevin explore tips, tricks, and techniques to get to market fast and cost-effectively with your GenAI powered MVP!

Phenomenal Cosmic Powers! Itty Bitty Living Space! Let's talk about Small Language Models

Joao explains the benefits of starting with small language models to keep costs manageable and ensure efficient scaling. He highlights that these models, characterized by a lower parameter count and memory requirements, can perform almost the same tasks as large language models but with improved operational efficiency. The discussion includes practical demonstrations, showcasing the deployment of these models on AWS Lambda, leveraging features like response streaming and function URLs for secure and efficient operations. This segment emphasizes the cost-effectiveness and scalability of using small language models, making them an ideal choice for startups looking to implement AI solutions without significant financial investment.
Watch the highlight on Twitch!

Ready to Go Prod? Serve Models with LangServe

We then transition into the practicalities of deploying AI applications to production, focusing on the use of LangServe. Joao introduces LangServe, a tool built on FastAPI, which allows for easy deployment of models in a production environment. This section aims to simplify the process of creating APIs for model interaction, particularly with AWS Bedrock integration.
Joao provides a detailed walkthrough of using LangServe, explaining how it supports role prompting to enhance model responses by assigning specific personas. He also highlights the flexibility of LangServe in testing and deploying various cloud models, enabling startups to create scalable and robust AI services. The discussion includes the use of LangChain AWS, a package that facilitates interaction with Bedrock, streamlining the deployment process. This segment is geared towards startups ready to transition from experimentation to full-scale production, offering tools and strategies to deploy models efficiently.
Watch the highlight on Twitch!

Serverless Retrieval Augmented Generation Powered by LanceDB

In the final segment, the focus shifts to the concept of Serverless Retrieval Augmented Generation (RAG) using LanceDB. Kevin and Giuseppe introduce RAG, a method that enhances AI applications by injecting relevant data into model responses, thereby improving their relevance and accuracy.
Chang She, CEO and co-founder of LanceDB, joins the discussion to provide insights into LanceDB's capabilities. He explains how this open-source vector database is designed for multimodal AI, offering efficient storage and retrieval of large datasets. The team demonstrates a serverless RAG application built using LanceDB and AWS Lambda, showcasing how it processes and retrieves data from S3 to generate accurate and contextual responses. Chang elaborates on the technical aspects, including disk-based storage, vector indexing, and integration with Apache Arrow, making LanceDB an ideal solution for scalable and efficient AI applications.
This section emphasizes the practicality and power of combining LanceDB with AWS services to build robust AI solutions, highlighting the seamless transition from local experimentation to cloud-based production environments.
Watch the highlight on Twitch!

Episode Resources

What's Next?!

In our next episode we’ll be talking about how Graph Databases are helping building the new generation of AI powered applications! Joining us we’ll have Camillo Anania, Head of Solutions Architecture for Healthcare startups here in the UK and Alex Bilbie, Head of Engineering at Muzz, the largest free dating app for Muslims. This is on July 4th, 3pm BST, 10am EST

Watch the Full Episode


Episode Engagement Metrics

Episode Engagement Metrics
Peak Viewers: 200
Average viewers: 70
Unique Chatters: 25
Messages: 116
CTAs as of 01/07: 139

Authors & Guests

Chang She is CEO and co-founder of LanceDB, a developer-friendly, open source database for AI: from hyper scalable vector search and advanced retrieval for RAG, to streaming training data and interactive exploration of large scale AI datasets! Follow Chang on LinkedIn.
João Galego is a Solutions Architect at Amazon Web Services, based out of Lisbon, Portugal. He has a background in Physics, a postgraduate degree in Forensic Studies and teaches a few classes at ISEG/ULisboa on applied AI/ML. When he's not working, studying or teaching, you can usually find him browsing the local bookstore, competing in hackathons or enjoying some quality family time. Follow João on LinkedIn.
Giuseppe Battista is a Senior Solutions Architect at Amazon Web Services. He leads soultions architecture for Early Stage Startups in UK and Ireland. He hosts the Twitch Show "Let's Build a Startup" on twitch.tv/aws and he's head of Unicorn's Den accelerator. Follow Giuseppe on LinkedIn
Kevin Shaffer-Morrison is a Senior Solutions Architect at Amazon Web Services. He's helped hundreds of startups get off the ground quickly and up into the cloud. Kevin focuses on helping the earliest stage of founders with code samples and Twitch live streams. Follow Kevin on LinkedIn

Any opinions in this post are those of the individual author and may not reflect the opinions of AWS.