RAG Architect: High-Throughput Knowledge Retrieval

Zoho; United States; Remote; USD 150,000 - 210,000; Full time

Job summary

Fyerx is seeking an experienced Context Window Optimization / RAG Architect to own enterprise generative AI retrieval performance, accuracy, and cost metrics. You will design high-throughput retrieval systems, optimize semantic parsing, build re-ranking pipelines, and engineer caching grids for data-grounded AI responses with minimal latency and maximum token efficiency.

As the lead, you will architect end-to-end RAG pipelines, optimize chunking and context usage, implement semantic caching, and

Qualifications

• 6 to 10 years of enterprise data engineering, database design, or search engine engineering experience.

Responsibilities

• Architect end-to-end advanced Retrieval-Augmented Generation (RAG) pipelines, with document parsing and multi-vector lookups.
• Optimize context window utilization with smart chunking, sentence-window retrievals, and sliding window strategies.
• Build high-performance re-ranking layers using cross-encoders (e.g., Cohere Rerank, BGE-Reranker).
• Implement automated semantic caching architectures to reduce API token costs and latency.
• Establish automated data chunking pipelines to parse PDFs, wikis, and SQL outputs into vector targets.
• Govern vector similarity spaces by fine-tuning hybrid search combining dense embeddings with BM25.
• Audit context-level hallucination rates and logs to improve precision and recall.

Skills

Python

Embedding models

Context window optimization

Performance engineering

Education

Cloud Data/Database Engineer Certification

Specialty Analytics credential (AWS/GCP/Azure)

Tools

Pinecone

Milvus

Weaviate

LlamaIndex

LangChain

BM25

GPTCache

Job description

Fyerx is seeking an experienced Context Window Optimization / RAG Architect to own enterprise generative AI retrieval performance, accuracy, and cost metrics. You will design high-throughput retrieval systems, optimize semantic parsing, build re-ranking pipelines, and engineer caching grids for data-grounded AI responses with minimal latency and maximum token efficiency.

As the lead, you will architect end-to-end RAG pipelines, optimize chunking and context usage, implement semantic caching, and