Project
Enabling AI inferencing on z/OS
Objective
To port, optimize, and deploy the open-source llama.cpp inference engine on IBM z/OS, enabling efficient AI model execution within enterprise workloads on the mainframe.
Outcome
. This effort aims to bring modern AI inference capabilities to z/OS, enabling seamless integration of machine learning models into enterprise workloads
| Apply By Date |
02 Jun 2025 |
| Students |
1 / 4 |
| Duration |
60 days |
| Mentor |
Haritha D |
Tools-Technologies | C |
Platform | 1 ) Mainframe Link to be provided once students start working on a project. |
College | | 1. PES University, Bangalore | | 2. PESIT |
|
Haritha D' Comment
- Environment Setup – 3 days
- Set up z/OS build environment for C++ development
- Identify and install required dependencies
- Adapt llamaccp for zopen – 2 days
- Analyze llama.cpp source code for system dependencies
- Adapt llamacpp for zopen framework
- Build & Compilation
- Resolve bootstrap issues – 7 days
- Resolve configure issues – 7 days
- Resolve build errors and validate successful compilation – 20 days
- Enable Unit test cases
- Enable unit. Tests and integrate test results to zopen framework – 7 days
- Validation
- Run sample models and validate output correctness – 3 days
- Benchmark performance – compare with zlinux and document results – 2 days
- Documentation
- Provide setup, build, and usage instructions for z/OS – 1 day
- Document any platform-specific changes made to the code – 3 days