We're a team of five computer science students planning our final-year project. We're comfortable with frontend and backend development and have basic AI/ML experience, although none of us are advanced in the field yet. We have about four months to research and prepare, followed by another four months of development. Since we'll also have other courses and exams, the project needs to be ambitious but manageable on a part-time schedule.
We'd like to build a substantial web application with meaningful backend and frontend components, while using AI/ML as an important part of the system rather than simply wrapping an existing AI service. Ideas such as malware detection and system-failure or anomaly prediction appeal to us because they combine machine learning with data processing, APIs, monitoring, investigation tools, and dashboards.
What are some realistic project ideas that would be large enough for five students, solve a genuine real-world problem, and fit within our preparation and development timeline?
3 Answers
You could build a project-planning and recommendation platform that helps users turn a broad goal into smaller tasks, suggests suitable technologies, and tracks progress. To make it more than a basic AI wrapper, include your own dataset, recommendation logic, user feedback loops, search and ranking, analytics, and a full collaboration dashboard. Just make sure the scope stays focused on one audience, such as students or small development teams.
Another possibility is a public-information analysis platform. It could collect articles and public reports from approved sources, group related events, detect recurring themes, compare claims with cited evidence, and present the results through timelines and visual summaries. The challenging parts would include source processing, document search, credibility indicators, explainable classification, and a well-designed web interface. You’d need to define the scope carefully and avoid presenting uncertain model results as established facts.
A strong option would be an incident-monitoring platform for logs and application metrics. It could ingest data from different sources, normalize and store it, detect unusual patterns, and display incidents on a dashboard. You could add investigation tools, alert prioritization, explanations for why an event was flagged, and an API for integrating with other systems. The machine-learning model would be one component of a broader product instead of the entire project.
That fits what we had in mind really well. I’ll add it to our shortlist.