I've built a small career platform and want to make it more than a standard CRUD application connected to an LLM API. It currently includes a resume builder with templates, drag-and-drop editing, live preview, PDF export, and AI-generated content. It also has a voice interview feature where a user provides a resume and job description, receives generated interview questions, and completes a spoken interview with basic scoring and feedback.
Because I plan to feature this project on my resume, I'm looking for genuinely challenging engineering improvements that would give me meaningful topics to discuss with a software engineering interviewer. The existing features can be approximated with general-purpose AI tools, so I'd like ideas that demonstrate deeper skills such as real-time systems, security, privacy, scalability, reliability, or thoughtful product engineering. I've already invested significant time in the project, so I'd prefer ways to evolve it rather than abandon it entirely.
4 Answers
You don’t necessarily need to invent a completely new product. A strong portfolio project can stand out by demonstrating measurable outcomes and thoughtful evaluation. For example, compare AI-generated interview feedback against a defined rubric, let users review or correct scores, track consistency across repeated interviews, and show how you detect low-confidence or unfair evaluations. Explain what problem the platform solves, how you measure improvement, and where the system deliberately avoids making unsupported claims.
If the project still feels too generic, narrow it around a specific user group or workflow rather than adding random features. You could support versioned resumes tailored to different job descriptions, explain why each change was suggested, maintain an audit trail, and add collaboration or review features. That gives you opportunities to discuss search, ranking, version control, authorization, background jobs, and data modeling while keeping the work you’ve already completed.
The feature list matters less than the engineering behind it. Treat the platform as a production system: isolate users properly, encrypt sensitive data at rest and in transit, avoid putting resumes or identifiers in logs, validate uploaded files and user-generated content, and design a reliable deletion workflow that removes personal data from databases, object storage, caches, search indexes, and backups where appropriate. Add tests, monitoring, structured documentation, and clear architectural boundaries so another developer could understand and extend the codebase.
That’s a useful direction. I’ll make privacy, deletion, and maintainability visible parts of the project instead of focusing only on adding more AI features.
For the voice interview, consider building a real-time audio pipeline instead of uploading raw recordings to an external service. A local voice-activity detector, streaming speech recognition, partial transcription, interruption handling, latency measurements, and resilient reconnect logic would create much more interesting engineering challenges. You could also explain how you handle noisy audio, backpressure, time limits, and the tradeoffs between local and hosted models.
That makes sense. A focused workflow with explainable suggestions sounds more valuable than simply adding another AI endpoint.