NVIDIA DAQIRI: реальный ИИ-инференс при высокоскоростном сборе данных

NVIDIA has introduced DAQIRI — a framework for embedding AI inference directly into a high-speed data collection pipeline. The goal: AI should work where data is born, not wait for it to be recorded and transmitted for processing.

Why Real-Time AI in the Stream Is Needed

When AlphaFold2 revolutionized drug discovery in 2020, its success depended entirely on ~170,000 protein structures collected by scientists since 1971 and stored in the Protein Data Bank. This is an illustrative example: good data is the foundation of any AI model.

But the classic "collect — save — train" approach has a fundamental flaw: there is a time lag between when data appears and when AI responds to it. In scientific experiments, industrial monitoring, or medical diagnostics, this time delay can cost the result.

What is DAQIRI

DAQIRI (Data AcQuisition with Intelligent Real-time Inference) — a software layer from NVIDIA that connects GPU-accelerated inference directly to a high-speed stream from instruments and sensors. The framework solves multiple tasks simultaneously:

• Reception and buffering of streaming data without loss
• Running AI models on GPU in real-time mode
• Filtering and tagging events on the fly — before writing to disk
• Integration with scientific instruments through standard interfaces
• Support for workflows where raw data volume exceeds storage throughput

The last point is particularly important: in physical experiments and genomic research, detectors generate terabytes per second. Recording everything is impossible — you must choose what to save. DAQIRI makes this choice in real time, using AI as a filter.

What Tasks Is This Relevant For

The framework is primarily aimed at scientific and industrial scenarios with high-speed data sources: particle accelerators and physics detectors, genome sequencing, industrial quality control on production lines, medical imaging in streaming mode.

"Measured data is the foundation of all AI models and workflows that

process data at the moment of creation, respond to what matters in real time, and analyze data for deep insights," — from the NVIDIA Developer blog.

In each of these cases, the value of an event diminishes over time. DAQIRI shifts AI as close as possible to the signal source.

What This Means

NVIDIA is consistently pushing the "AI everywhere" concept — not just in data centers, but at the edge of infrastructure, right in measurement instruments. DAQIRI is another step in this direction: inference moves to where data originates, not where it's convenient to store.