AnnaSuSu/TechSpar

What it solves

TechSpar addresses the fragmented nature of technical interview preparation. Instead of simply generating a list of questions, it creates a continuous feedback loop where training, resume analysis, job description (JD) alignment, and real-time assistance are integrated into a single system that tracks a user's progress and weaknesses over time.

How it works

The system builds a long-term user profile and knowledge base to personalize the experience. It utilizes several AI-driven modules:

• Training & Simulation: Dynamically generates questions based on a question bank, the user's historical weak points, and mastery levels. It can simulate interviews based on a resume or specific job descriptions.
• Real-time Copilot: Provides live ASR (Automatic Speech Recognition), predicts follow-up questions, suggests answers, and flags risks during an actual interview.
• Review & Analysis: Transcribes recordings into structured Q&A and provides detailed analysis and improvement suggestions for each answer.
• Knowledge Management: Allows users to import PDFs, DOCX, and Markdown files to provide context for the AI agents.
• Scheduling: Uses the SM-2 algorithm to schedule reviews of weak areas.

Who it’s for

Engineers and technical professionals preparing for job interviews who want a structured, data-driven way to track their improvement and identify knowledge gaps.

Highlights

• Integrated Loop: Connects specialized training, resume-based mocks, and JD-specific prep into one evolving profile.
• Real-time Assistance: Offers a live Copilot with speech-to-text and answer suggestions.
• Local-First Options: Supports self-hosting via Docker and local embeddings using Transformers.js and ONNX to avoid external API dependencies for vectorization.
• Flexible Deployment: Available as a web app, a desktop client (Electron), or a self-hosted instance.