Ghost Mentor Match: Channeling Coding Spirits & a Git Ofrenda with Qwen & Gemma

Description and use

Ghost Mentor Match offers privacy‑preserving, offline AI mentorship and a digital altar for deleted git commits, leveraging open‑weight models Qwen and Gemma.

Problem

Developers need a privacy‑preserving, offline‑capable mentorship tool that can analyze stack traces, architectural roadblocks, and historical git commits without exposing proprietary code to external APIs.

Outcome

The platform delivers privacy‑preserving, offline mentorship and a unique code‑ofrenda experience, demonstrating the benefits of open‑weight models for developer tooling.

Workflow

• 1. User inputs tech stack, stack trace, or git log into the web UI.
• 2. The system calculates compatibility scores and matches the user with a specialized AI mentor persona.
• 3. The mentor diagnoses the issue and provides concrete fixes, presented as terminal poetry.
• 4. For git logs, the model reads the diff context and generates a heartfelt tribute on a digital altar.
• 5. Users can toggle between cloud‑hosted Qwen and local Gemma inference via a single UI switch.

Evidence

The primary reasoning and persona‑matching engine is powered by the open‑weight Qwen 3.8‑27B model running on ultra‑fast Cerebras Cloud inference, paired with a native local inference fallback using Google’s Gemma via Ollama

Source

Running Gemma locally or routing through self‑controlled open weights ensures sensitive source code isn’t harvested to train a closed vendor’s next commercial model

Source

By supporting local open weights on consumer hardware, developers can summon mentor diagnostics completely offline at $0.00 token cost

Source