Scientific Computing & AI

Our capabilities span physics-informed neural networks (PINNs), machine learning and inverse design, neuromorphic computing, and processing-in-memory (PiM) architectures for scientific and edge-computing applications. P3P develops and deploys AI models for problems including materials optimization, nuclear instrumentation, autonomous data analysis, and computationally efficient inference.

We also maintain private, on-premises AI infrastructure, including a locally deployed small language model (SLM) environment that enables secure development of domain-specific scientific AI tools without requiring sensitive technical data to leave P3P-controlled systems. Our hardware research includes FPGA-based acceleration, spiking neural networks, and emerging ReRAM-based PiM computing, allowing us to investigate algorithms and computing architectures together rather than treating software and hardware independently.

(Expanded technical content is currently being prepared alongside
active research programs and will be added soon.)