RAG Developer

About the Role

We're seeking a RAG Developer to build retrieval-augmented generation systems on GCP. You'll design RAG pipelines using AlloyDB, PgVector, and LLM frameworks—from document chunking and embeddings to evaluation and deployment.

Key Responsibilities

• Design and implement RAG architecture leveraging AlloyDB and PgVector

• Develop document chunking strategies (semantic, hierarchical, sliding window, recursive, token-aware)

• Build and optimize vector embedding pipelines

• Integrate LLMs with retrieval systems using LangChain, LlamaIndex, or Vertex AI

• Design evaluation frameworks (RAGAS, retrieval metrics, generation quality)

• Optimize vector search, indexing, and query performance

• Monitor, debug, and improve production RAG systems

Required Qualifications

• 3+ years software development experience

• Hands-on experience with AlloyDB, PostgreSQL, and PgVector

• Strong knowledge of vector embeddings and embedding models

• Expertise in document chunking methodologies and their impact on retrieval

• Experience building RAG systems (LangChain, LlamaIndex, or similar)

• Proficiency in Python

• Experience with GCP services (Vertex AI, Cloud Run, Cloud Functions, Compute Engine)

• Understanding of LLM APIs and integration patterns

• Knowledge of SQL optimization and information retrieval principles (TF-IDF, BM25)

Preferred Qualifications

• Experience with other vector databases (PgVector, AlloyDB, Qdrant)

• Advanced RAG patterns: hybrid search, re-ranking

• Knowledge of knowledge graphs and LLM safety/hallucination mitigation

• Experience with evaluation frameworks (RAGAS, custom metrics)

• Background in NLP, machine learning, or information retrieval

Core Technical Skills

Category Technologies

Vector DB AlloyDB, PgVector, Qdrant

Embeddings OpenAI, Google text-embedding, HuggingFace, Cohere

RAG Frameworks LangChain, LlamaIndex, Semantic Kernel, Prompt Flow, RAGAS

GCP Services Vertex AI, Cloud Run, Cloud Functions, Compute Engine

Languages Python, SQL

DevOps Docker, Git, CI/CD, Cloud IAM, Monitoring

Key Expertise Areas

Document Chunking

Semantic, Hierarchical, sliding window, recursive, and token-aware chunking strategies; metadata preservation

Vector Database Optimization

Index optimization (HNSW, IVFFlat, ScaNN), similarity search algorithms, query performance tuning, hybrid search

RAG Architecture

End-to-end system design, query processing, relevance ranking, context management, prompt engineering

Evaluation & Monitoring

Retrieval metrics (Precision@k, Recall, MRR, NDCG), generation quality metrics (ROUGE, semantic similarity), production monitoring