RAG Knowledge Base

Retrieval-Augmented Generation pipeline for intelligent interview context

0
Total Embeddings
0
Source Types
3-small
Embedding Model
On
HyDE Search
Indexed Content by Source Type
Resumes
0 chunks
Job Roles
0 chunks
Questions
0 chunks
Transcripts
0 chunks
Async Video
0 chunks
Knowledge
0 chunks
Companies
0 chunks
ATS Jobs
0 chunks
ATS Candidates
0 chunks
AI Agents
0 chunks
Pipeline Config
Chunk Size 500 chars
Chunk Overlap 50 chars
Top-K Results 5
Re-rank Top-K 3
Max Context 2000 tokens
HyDE Enabled
Embedding text-embedding-3-small
RAG Pipeline
Document Ingestion
10 source types: Resumes, Jobs, Transcripts, Videos, Knowledge & more
Chunking
Sentence-boundary aware, 500 char chunks
Embedding
OpenAI text-embedding-3-small
pgvector Storage
Cosine similarity search
HyDE + Re-rank
Hypothetical docs, source-type boosting