Build an agentic AI module that transforms a full software repository into an executable HPC workload on Slurm and outputs a Workload Execution Requirements (WER) profile for CloudLift.

Responsibilities

• Analyze repositories (source files, notebooks, scripts, configs, build/dependency files, docs)
• Extract software requirements (Python packages, system libs, compilers, MPI/OpenMP/CUDA, modules, containers, external data)
• Infer build/run instructions from artifacts (README, requirements.txt, pyproject.toml, setup.py, Makefile, CMakeLists.txt, Dockerfile, shell scripts, notebooks, CI/CD workflows)
• Produce a safe action plan: clone, checkout, install, build, test, execute, collect logs, extract workload requirements
• Generate Slurm scripts tailored to workload type (CPU/GPU, memory, time, modules, env setup, command)
• Execute safely, monitor jobs, collect stdout/stderr, scheduler metadata, runtime/memory/GPU metrics
• Diagnose failures (missing deps, wrong paths, failed builds, missing datasets, permissions, invalid Slurm opts, CPU/GPU mismatches)
• Enable codebase Q&A for users (what it does, how to run, deps, GPU needs, failure reasons)
• Produce final WER profile and convert to CloudLift input format

Required skills

• Strong Python; Linux/Git/shell familiarity
• Basics of pip/conda/make/CMake/Docker
• Interest in HPC, Slurm, workload execution; agentic AI/LLM tool use/structured reasoning/validation
• Work with JSON/YAML and clear technical documentation

Expected outcome

• A working prototype linking repository understanding with HPC execution and CloudLift recommendation.