Kannow77/somali-ai-interview-coach

A Somali-first AI interview practice coach built with Gradio and Hugging Face Inference Providers. --- title: Somali AI Interview Coach emoji: 🤝 colorFrom: indigo colorTo: green sdk: gradio python_version: "3.12" app_file: app.py suggested_hardware: cpu-basic pinned: false --- # Somali AI Interview Coach An interview-practice application for Data Science, AI Engineering, and Software Engineering candidates. It supports Somali, English, and bilingual feedback, and turns a short practice session into structured, actionable coaching. The application is built with Gradio and Hugging Face Inference Providers. It is designed for learning and interview preparation only. It must not be used as an automated hiring, rejection, ranking, or employment decision system. This repository integrates a hosted language model; it does not train or ship a custom model. ## What it does 1. A learner selects a career track, experience level, and feedback language. 2. The application presents a curated technical interview question. 3. The learner submits an answer in Somali, English, or a mixture of both. 4. A Hugging Face language model evaluates the answer against a structured feedback contract. 5. The learner receives a score, strengths, missing concepts, an improvement suggestion, and a focused follow-up question. 6. The current browser session displays the interview history so the learner can review progress. ## Features | Area | Included capability | | --- | --- | | Career tracks | Data Science, AI Engineer, Software Engineering | | Experience levels | Junior, Mid-level, Senior | | Starter questions | 27 curated questions: three for every track and level combination | | Languages | Somali, English, Somali + English | | Feedback | Score, summary, strengths, missing concepts, improved answer, follow-up | | AI integration | Hugging Face InferenceClient with structured JSON output | | Reliability | Safe parsing and an explicitly labelled offline demo fallback | | Privacy by default | No database or answer persistence is implemented | | Deployment | Ready for a CPU-based Hugging …