Advice - GSoC 2026: OpenVINO Deep Search AI Assistant on Multimodal Personal Database for AIPC

[SudharshanamBuddha]

Hi everyone,

I’m Sudharshanam, a Computer Science graduate from India, interested in contributing to OpenVINO for GSoC 2026, particularly in areas related to model inference, optimization, and system-level AI workflows.

I’ve recently started exploring OpenVINO’s architecture and have been going through operator-level discussions (such as TopK and NaN handling), as well as understanding how different components of the inference pipeline interact across CPU and GPU backends.

I’m especially interested in how efficient inference systems can be built by combining model optimization, hardware-aware execution, and intelligent data handling. I’m also exploring ideas around retrieval-based systems and how OpenVINO can be leveraged in such pipelines for real-world AI applications.

From my initial exploration, I’m trying to better understand:

• How operator implementations are structured and extended across versions
• How consistency is maintained across different backends (CPU, GPU)
• Where contributors can start adding value in terms of features, testing, or performance improvements

I have experience with Python, full-stack development, and I’m currently focusing on AI/ML systems, including projects involving automation and data processing. I’m now aiming to deepen my understanding of ML infrastructure and contribute meaningfully to projects like OpenVINO.

I’d really appreciate any guidance on:

• Good starting points for contributors at the operator/runtime level
• Areas where help is currently needed (issues, testing, documentation, etc.)
• Any recommended resources to better understand OpenVINO internals

Looking forward to learning from and contributing to the community!