If 42 is the ultimate answer, what’s the ultimate question?

Well, unfortunately, I can’t help there.

But, if you want to know more about questions and answers, specifically developing a Q&A system using LLMs, you’re in luck!

On Jan 17th Rahul’s hosting a half-day hands-on workshop in San Francisco. You’ll create a proof-of-concept prototype, and in the process learn how to create a data pipeline for preprocessing, ingest data into a vector database, semantic search for the answer from the question, and compile a response for the user using an LLM.

It’s a paid workshop, and you can find more info and purchase it here.

For those not able to attend, you can still tap into Rahul’s rich RAG repository by buying the remote course (with updated source code!) here for the knocked-down price of $149, before the discount ends at the end of the month.

Not quite the answer to life, the universe and everything, but a step closer at least.

Language, Graphs, and AI in Industry // Paco Nathan // MLOps Podcast #201

Sometimes a tech tip can be called a ‘secret sauce’.

Well, at the start of this episode I mention my actual (not-so) secret sauce (Valentina) and Paco, was kind enough to email this recipe as a follow up, so I thought I’d share it with you all too!

Now we’ve got our snacks sorted, lets get down to it!

This was a great chat with a real luminary in the field of AI. He’s so knowledgeable, telling me all about the Macy conferences in the late 40s and 50s and the key players in those. Plus his own experiences with neural networks, involvement with Apache Spark, and being one of the pioneer users of AWS for Hadoop clusters.

We get in to the trend of moving away from broad, one-size-fits-all AI models to more specialized models, and the need for efficiency and specificity.

And we talked manufacturing, which was really interesting. I had in my mind, like others, an idea of how ML would be used. But, no! Paco told me how AI’s really making huge impacts in decision-making and efficiency in manufacturing environments. And it may not be how you're thinking.

So, sit back and enjoy your snacks and the podcast!

Small Data, Big Impact: The Story Behind DuckDB // Hannes Mühleisen and Jordan Tigani // MLOps Podcast #202

When it came to bigger and bigger data, Hannes just said Duck No!

He’d had bad experiences and knew thought there must be a better way, leading to the creation of Duck DB, an open-source, high-performance, embedded SQL database. Hannes discusses the journey to its development, emphasizing the commitment to enhancing the developer experience – an aspect so well-received that it borders on tech evangelism. Interestingly, he mentions that their focus on the user experience was so intense that they haven't even begun optimizing the database yet.

We also talk about the unique relationship with MotherDuck. Jordan views this collaboration as a potential blueprint for others. This unconventional partnership highlights the decisions they’ve made that go against the norm. Another being to opt for a single-node architecture for Duck DB. Bold choices, but ones which have since proven to be wise choices.

So, is it worth a listen? Duck yeah!

Join us for a discussion surrounding Kolena's latest research on hallucination detection for Large Language Models (LLMs).Gain insights on:

• Methodologies, challenges, and advancements in the field
• Hands-on perspectives on the complexities and solutions of LLM testing
• A compare and contrast of leading models
  
  

Learn meaningful implications for evaluating the performance of LLM models and reducing the negative impact of hallucinations in production!

Register now to save your seat!

Plus, all registrants will be sent a recording of the webinar.

ML-augmented R&D in Biotechnology // Lucile Bonnin // IRL Meetup #58 Berlin

Faster, cheaper and better.

No, not a Daft Punk song, but rather Cambrian’s ‘very simple challenge’ on how to do things to address the environmental impact of material production through biotechnology.

Lucille explains Cambrium's use of AI and machine learning to create sustainable materials, replacing traditional ones like concrete and plastics. She details how Cambrium uses genetic AI for protein design and machine learning for scaling, focusing on their first product, Novacol, for skincare and beauty.

The talk also covers machine learning integration in Cambrium's R&D, from protein design to product scale-up, and hints at future innovations like five-letter DNA sequences and new testing protocols.

Fascinating work that’ll surely have a positive impact Around The World.

Watch it here!

💡Job of the week

Data Scientist // DataThings (Luxembourg)

1. Analyze client/project data for analytics and machine learning solutions.
2. Conduct experimental studies for approach accuracy.
3. Participate in product integration and quality testing.

1. Masters or engineering degree in a relevant field.
2. Expertise in machine learning algorithms and data processing.
3. Programming skills in Python, C/Java, and machine learning frameworks.

How to Build Your First Semantic Search System: My Step-By-Step Guide with Code

Big data’s great, but sometimes finding what you need is like a game of Where’s Waldo? Just not as fun.

This handy walk-through demystifies the process by walking readers through the development of a semantic search engine for a dataset of research papers.

Leveraging tools like Milvus-lite vector database and Cohere Embed for text embeddings, the guide lays out a clear path from conceptualization to execution. It not only discusses the technicalities of setting up a vector database and embedding data but also touches upon fine-tuning and optimizing search results.

Whether you're a novice or an experienced practitioner in the field of machine learning and data science, this comprehensive guide, complete with code snippets, is a must-read.

Plus, you don’t have to search for it, just click the link!

With thanks to Sonam Gupta for the contribution.

Stockholm - January 16
Copenhagen - January 17
London - January 18
Stockholm - January 18
Madrid - January 25
Frankfurt - January 25
Edinburgh - January 25
Seattle - January 26
Munich - January 30
Denver - January 30