How Sapu Indexed 28 Million PubMed Abstracts to Accelerate Cancer Research with Qdrant

2026-05-12 · Source: Qdrant Blog on Qdrant - Vector Search Engine · Field: Science & Research — Health & Medical Research, Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Intermediate, medium

Summary

Sapu, an early-stage biopharmaceutical company developing nanomedicine treatments for hard-to-treat cancers, leverages an internal AI platform to accelerate research. Initially encountering stability issues with self-hosted vector search, Sapu transitioned to Qdrant Cloud Premium for enhanced reliability, SOC 2 compliance, and superior support. This strategic move enabled the company to index 28 million PubMed abstracts into a single Qdrant collection, making the entire biomedical literature corpus instantly searchable. The platform, which utilizes OpenAI's text-embedding-3-large model at 3,072 dimensions and LangChain, supports hybrid retrieval, metadata filtering, and a novel per-document parallel querying pattern. This infrastructure has facilitated the publication of seven peer-reviewed research papers and expanded into partnerships, with future plans including Qdrant Edge for air-gapped deployments and Qdrant Cloud Inference.

Key takeaway

For AI Engineers building scalable retrieval systems in regulated industries, consider managed vector database services like Qdrant Cloud Premium. Your focus should remain on core research and application development, not infrastructure management. This approach ensures reliability, simplifies compliance with standards like SOC 2, and provides access to expert support and continuous feature updates, accelerating your path from prototype to production-grade AI solutions.

Key insights

Scaling biomedical research with AI requires robust, compliant vector search infrastructure like Qdrant Cloud.

Principles

• Prioritize managed services for critical infrastructure.
• Compliance certifications are crucial for platform licensing.
• Hybrid retrieval enhances search precision.

Method

Sapu's pipeline uses OpenAI's text-embedding-3-large (3,072D) with LangChain for ingestion, writing to Qdrant Cloud for vector and metadata-filtered search, and a separate SQL DB for access control.

In practice

• Index entire public datasets for comprehensive search.
• Implement per-document parallel querying for deep literature review.
• Use payload filters for granular access control.

Topics

• Vector Search
• Qdrant Cloud
• Biomedical Research
• AI Platforms
• PubMed Database
• SOC 2 Compliance
• Hybrid Retrieval

Best for: AI Engineer, Machine Learning Engineer, Research Scientist

Related on AIssential

• Qdrant Beats Elastic’s DiskBBQ at 2x Throughput, Half the Latency, and 1/3 the Compute — Proper configuration of vector databases is crucial for optimal disk-based search performance and resource efficiency.
• How a Nonprofit Transforms Data with Cloudera and AI — Nonprofits can leverage hybrid data and AI platforms to conduct complex research affordably.
• Building a High-Performance Vector Search Engine from Scratch in 2026 — Building a custom vector search engine offers maximum privacy and zero latency overhead for AI applications.
• Vector Database Shootout: Postgres+pgvector vs. Qdrant vs. Pinecone — Vector database selection for RAG now prioritizes scalability and cost-efficiency over basic functionality.
• Now Available on Qdrant Cloud: GPU Indexing, Multi-AZ, and Audit Logging — Qdrant Cloud enhances enterprise readiness with GPU-accelerated indexing, Multi-AZ replication, and comprehensive audit logging.
• Deploy Qdrant on Kubernetes: A Production Ready Step-by-Step Guide — Deploying Qdrant on Kubernetes requires careful configuration of namespaces, secrets, configmaps, services, and statefulsets for production readiness.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Qdrant Blog on Qdrant - Vector Search Engine.