Data Science Dojo & Weaviate
AI in the
Real world
Victoria Slocum
Machine Learning Engineer
webinar #1
Each number represents how much red, green, or blue is in the color.
This is exactly what a vector embedding is - a sequence of numbers that represents meaning.
so...
how do you actually choose an embedding model?
Data Performance
Infrastructure considerations
Language Specificity
domain Specificity
comparison/
benchmarking
inference costs
storage costs
latency /
throughput
ML is full of trade-offs
webinar #2
search systems
run the world
Search systems are everywhere,
from Google, to e-commerce sites, to internal documents
basic Vector search is just* math
the weird ml part is all in the vector embeddings
So...
why do you even need a vector database?
why do you even need a vector database?
ANN
kNN
what does at scale mean?
The difference between...
AI apps
in the real world
AI apps
in the real world
1
Search at scale
2
Retreival augmented generation
3
ecommercewith ai
4
agentic workflows
Search at scale
flexible deployment & architecture
Hybrid search that scales
customizability without sacrificing usability
community, enablement, and education
flexible deployment & architecture
Hybrid search that scales
customizability without sacrificing usability
community, enablement, and education
Multi-tenant systems
flexible deployment & architecture
Hybrid search that scales
customizability without sacrificing usability
community, enablement, and education
flexible deployment & architecture
Hybrid search that scales
customizability without sacrificing usability
community, enablement, and education
the balance between
out-of-the-box defaults
while still being able to build custom solutions
flexible deployment & architecture
customizability without sacrificing usability
community, enablement, and education
open source 💙
Retreival Augmented Generation
(RAG)
The next generation
of RAG
Follow to stay updated on the release!
EDWARD
danny
E-commerce with AI
From the beginning, Wood and Reed’s vision was to use the power of AI to help shoppers more easily discover the products they love.
E-commerce is the intersection of a lot of different vector search strengths - and difficulties
- Multi-modal embeddings
- Large-scale, single-tenant search
- Recommendation and personalization systems
- Prioritize speed over exact accuracy
- Query transformation
Agents
The Next Horizon
Agents
The Next Horizon
What is an agent?
- LLM: Acts as the brain of the operation.
- Tools: External resources the agent can access.
- Memory: Both short-term and long-term.
- Reasoning: Break down problems and plan solutions
An unsolicited opinion:
Purpose-built agents with some degree of human-in-the-loop will be the next big thing we see in industry, from there, we can move onto more autonomous systems (but this will take awhile)
Agentic query augmentation
for example...
- User input passed to LLM
- LLM decides which collection to query, or tool to use
- LLM designs a query with filters, aggregation, etc.
- Search + Generation is performed
Agentic query augmentation
for example...
like in Elysia
Agentic query augmentation
for example...
purpose build agents
