🎉 Weaviate Monthly Introductions - May 2023

[dandv]

Hello everyone and welcome to the “Weaviate Monthly Introductions” thread!

If you’ve just joined our community, we want to extend our warmest welcome. This thread is the perfect opportunity to introduce yourself, get to know the other members, and to learn more about how diverse and vibrant our Weaviate community is.

To start things off, please tell us a little about yourself and how you are currently using or planning to use Weaviate. Here are some questions for inspiration:

• What’s your name, location, and what you do? (only share what you’re comfortable with, of course!)
• How did you find out about Weaviate?
• What are you currently using Weaviate for? Could you tell us a little about your project?
• Are there any specific features of Weaviate that particularly impressed you or drew you in?
• What’s one challenge or problem you’ve solved using Weaviate? Or perhaps you’re facing a challenge that you hope Weaviate will help you solve?
• If you could improve or add one feature to Weaviate, what would it be?
• Any fun facts about yourself? (Optional, but encouraged!)

Let’s keep this thread friendly, welcoming, and supportive. Remember, there’s no such thing as a “stupid” question, so don’t hesitate to ask if you’re not sure about something!

Also, experienced members, don’t shy away from introducing yourselves again, your experiences could be invaluable to our new members. Feel free to share some advice or insights you’ve gained along the way.

We’re thrilled to have you as part of our community and can’t wait to hear from you!

Cheers,
Dan

P.S. Don’t forget to check back often to welcome new members. And let’s keep the conversations going!

[etiennedi]

Uhh, let me kick this off.

@dandv

• What’s your name, location, and what you do? (only share what you’re comfortable with, of course!)

Etienne, , Weaviate Co-Founder & CTO

@dandv

• How did you find out about Weaviate?

Bob was pitching Weaviate (back then a “knowledge graph”) to a company that I freelanced for. We met there, the rest is history. A loooot has changed since then. No more knowledge graph

@dandv

• What are you currently using Weaviate for? Could you tell us a little about your project?

The two most recent use cases were load test for large-scale multitenancy cases (v1.20 coming out in about a month ), and performance tests on various GCP & AWS infrastructure. Some big surprises there, could make for a cool blog post at some point.

@dandv

• Are there any specific features of Weaviate that particularly impressed you or drew you in?

At the moment I’m most excited about everything related to multi-tenancy and the massive potential we have there for separation of storage and compute

@dandv

• What’s one challenge or problem you’ve solved using Weaviate? Or perhaps you’re facing a challenge that you hope Weaviate will help you solve?

Affordable, massive-scale, multi-tenant, highly concurrent, vector + hybrid search use cases

@dandv

• If you could improve or add one feature to Weaviate, what would it be?

My pet peeve is that we have three APIs (REST, GraphQL, gRPC), but there is no feature parity between them. You need to use at least two different APIs for an e2e use case.

@dandv

• Any fun facts about yourself? (Optional, but encouraged!)

I’m a big wine geek and my demos and example cases often revolve around wine. It’s a fun topic because it has a very domain-specific language yet is still relatable to a large audience.

[kureev]

• What’s your name, location, and what you do? (only share what you’re comfortable with, of course!)

Hi everyone! My name is Alexey Kureev, founding engineer @ Novoic (UK). I help very smart people to build software for early Alzheimer’s disease (and MCI) diagnostics.

• How did you find out about Weaviate?

Frankly speaking, I discovered Weaviate once I’ve been searching for a vector db to use for my pet project. I’ve tried Chroma first, then moved to Pinecone and settled down with Weaviate

• What are you currently using Weaviate for? Could you tell us a little about your project?

I build a pet project that revolves around the idea of contextualising LLM for developers. For example, I want this tool to be able to help with tests, but even this proves to be a non-trivial task as it might seem from the first glance. Copy-pasting code into LLM just doesn’t cut it: it doesn’t know about existing mocks, libraries you use, downstream dependencies of your code etc etc. With proper contextualisation, I believe it will be possible to solve all these problems and significantly improve DX.

Weaviate helps me to:

• Structure my context (using different classes with cross-references changed my life)
• Reduce complexity on my side by delegating vectorization to text2vec modules
• Hybrid search for code references and other information in the codebase

• Are there any specific features of Weaviate that particularly impressed you or drew you in?

Cross-referencing. Modules.

• What’s one challenge or problem you’ve solved using Weaviate? Or perhaps you’re facing a challenge that you hope Weaviate will help you solve?

I’m somewhere in the middle. The main challenge is to analyse code and not only vectorise contents of the codebase, but also draw some conclusions about the structure and files that can help answering generic questions. Maybe, pre-generate an “information” object with the summary of the findings, idk yet.

• If you could improve or add one feature to Weaviate, what would it be?

Frankly speaking, documentation. Some things are outdated, some are not elaborative enough. For cross-referencing, I had to jump between three different topics as the information scattered around the entire docs website . Would be great to have more domain-oriented structure with solid examples etc.

• Any fun facts about yourself? (Optional, but encouraged!)

Not sure how fun of a fact it is, but I studied to be a rocket scientist, so usually I am the most qualified person to say “it’s not rocket science”

[david]

• What’s your name, location, and what you do? (only share what you’re comfortable with, of course!)

Hey there! I’m David, software engineer / tech entrepreneur. I’m the founder of inRecovery, an addiction recovery platform that helps people overcome addiction and thrive in life.

After my 2nd tech exit, I struggled many years with drugs and my mental health. Eventually, I went to rehab and saw a major need for innovation. inRecovery was born from a vision to make care more data-driven, personalized, and create a more connected recovery experience for patients and their families.

• How did you find out about Weaviate?

I’ve been pretty involved in Open Source since I can remember and part of the Apollo GraphQL community. So I found Weaviate while researching a vector database solution that could fit our needs. But more importantly that has the right team, culture, and community behind it.

• What are you currently using Weaviate for? Could you tell us a little about your project?

We are building context vectors for Providers from their knowledge bases/historical data. As well as individual patient’s data like treatment plans, notes, lab results, progress, etc. The ultimate goal is to provide a NLQ interface for both clinicians and patients to improve their decision making and provide support, respectively.

• Are there any specific features of Weaviate that particularly impressed you or drew you in?

Mostly the community. When I saw the Slack channel, the enthusiam of @bobvanluijt , Conner, @etiennedi, @erika-cardenas

Btw @etiennedi , I LOVED your FOSEM talk on building in Go. From optimizing memory allocation using buffer (binary.LittleEndian.Uinit32), choosing binary encoding over JSON, and looping over raw data / skipping decoding. So cool to see your process, I’ve shared this with many developers as a great approach to problem solving.

• What’s one challenge or problem you’ve solved using Weaviate? Or perhaps you’re facing a challenge that you hope Weaviate will help you solve?

We have unique challenges in the healthcare space. Mostly around privacy and the need for strong semantic layers given the sensitivity of our domain.

• If you could improve or add one feature to Weaviate, what would it be?

A better and faster onboarding experience. I have some idea here and I’d love to contribute our UX and dev resources to make it happen. We should connect!

• Any fun facts about yourself? (Optional, but encouraged!)

• I was expelled from my High School for hacking my school district
• I’ve always loved engineering, but wanted to study art
• My dream is to raise Alpacas with my girl, somewhere remote, surrounded by nature
  

[jphwang]

Hi @david

Wow, what a great project and such an amazing story behind it. Congratulations on both, but especially on your recovery.

I also noticed this:

@david

A better and faster onboarding experience. I have some idea here and I’d love to contribute our UX and dev resources to make it happen. We should connect!

This is an area that we’re actively looking at - I’d love to get a bit of your time to connect & have a chat. Would you be open to that? Or written communication is fine too :).

If that works, please drop me a DM, we would love to get your feedback on your impressions so far, and talk about how we could improve things.

Cheers!
JP

(Alpacas are fab. There are some near my in-laws’ place and we love walking/driving by them. So fluffy.)

[jphwang]

Nice to see you here @kureev.

Sorry about your experience with the cross-refs. Actually, @dandv and has put this page together in the last few days - Cross-references | Weaviate - vector database.

Obviously it would have been better to have had it earlier, but we’d love to get your thoughts on it. Pls let me / Dan know if you have thoughts on it, good or bad.

We are building up a library of task-oriented, “how-to” sections in our docs to address exactly this. We don’t get there all at once, but we’re listening & trying our best.

[etiennedi]

@david

Btw @etiennedi , I LOVED your FOSEM talk on building in Go. From optimizing memory allocation using buffer (binary.LittleEndian.Uinit32), choosing binary encoding over JSON, and looping over raw data / skipping decoding. So cool to see your process, I’ve shared this with many developers as a great approach to problem solving.

Thank you Glad to hear that it was useful.

[kureev]

Oh, that’s very nice! The article is indeed what I’ve been looking for some time back. One suggestion: it might be beneficial to consider adding a section about batch referencing. I had a case where I need to create plenty of cross-references at once. Maybe a reference to the batch update docs would be enough, though.