Dmitry Petrov

Text-to-SQL makes great demos, but in real systems generating queries is rarely the hard part - understanding data is. Modern data is increasingly S3-first and multimodal, where meaning is defined by Python workflows, not table schemas.

AI systems often need just a slice of a large file – a clip from a video, a segment of audio, or a time-windowed spectrogram – not the entire thing. This talk explores real-world use-cases for querying inside video, audio, and sensor data stored in cloud storage – enabling tasks like segmentation, object detection, speaker filtering, and targeted, context-aware LLM inputs.

The team behind DVC has spent years tackling data versioning challenges. With the rise of AI, we’ve seen new complexities emerge - especially with multimodal datasets like images, video, audio, and text.

AI and ML are becoming an essential part of software engineering. Open source tools like Git, Git-LFS, MlFlow can increase ML teams productivity by introducing best practices.

Python is a prevalent programming language in machine learning (ML) community. A lot of Python engineers and data scientists feel the lack of engineering practices like versioning large datasets and ML models, and the lack of reproducibility.