Piyush Ratan Logo
    HomeWorkToolsAboutContact
    About MeView Tools
    All Work

    I built the phishing classifier, scoring flow, and Flask backend

    CYBERALERT

    An AI-assisted spam and phishing detector that classifies suspicious messages, scores severity, and explains risk in plain language.

    CYBERALERT project visual

    Project Type

    Spam and phishing detector

    Stack

    Python, Flask, AI APIs, prompt engineering

    Core Work

    Message classification, severity scoring, risk explanation

    Timeline

    Built in 2026

    Case Study

    Engineering Notes

    01

    Project Overview

    CyberAlert is a prototype for detecting spam and phishing messages. It is built for users who receive suspicious links, offers, login warnings, or payment messages and need a quick technical explanation without reading a cybersecurity textbook first.

    02

    Problem / Motivation

    Most people can sense that a message feels wrong, but they may not know which signal matters: urgency, spoofed sender, credential request, bad link, or emotional pressure. A yes/no detector is too shallow because it does not teach the user what to look for next time.

    03

    Architecture / System Design

    The backend is built with Python and Flask. A submitted message moves through request validation, prompt construction, AI analysis, severity scoring, and a cleaned response payload. The prompt asks the model to classify the threat type, explain the indicators, and return a practical risk level.

    The system is intentionally lightweight because the important part is the classification flow, not a giant backend. If the prompt is loose, the output gets vague. So the prompt structure carries a lot of the reliability work here.

    04

    Key Features

    CyberAlert is focused on explainable risk detection.

    • Spam and phishing classification for submitted messages.
    • Severity scoring for prioritizing risky content.
    • Plain-language explanation of suspicious indicators.
    • Prompt-engineered response structure for consistency.
    • Flask API flow that can be extended into a larger security tool.

    05

    Technical Challenges

    The biggest issue is consistency. AI can over-explain low-risk text or under-explain subtle phishing attempts if the prompt does not define a clear output shape. Security tools also need careful wording because a confident false safe result is worse than an honest uncertain one.

    06

    Solutions / Engineering Decisions

    I chose Flask because the prototype needed a direct request-response backend with low overhead. The AI prompt is structured around threat category, severity, evidence, and recommendation so the result stays readable. The system favors cautious explanations over dramatic claims.

    07

    Outcome / Final State

    The prototype can analyze suspicious messages and produce a clear risk explanation. It is a practical foundation for a bigger security workflow with URL checks, sender validation, and stored reports added later.

    CybersecurityFlaskAIPrompt EngineeringPython

    Key Capabilities

    Classifies spam and phishing messages with an AI-assisted analysis flow.

    Scores severity so risky messages can be prioritized instead of treated equally.

    Uses prompt engineering to make the reasoning clearer for users.

    Links

    GitHub

    Seedha kaam. What I built, how it runs, what stack I used, where the control sits.

    More

    Keep Moving

    All Work

    2026

    PCM Syllabus Tracker

    A brutalist, high-performance spaced repetition tracker for Class 11 and 12 Physics, Chemistry, and Mathematics (PCM).

    2026

    DARPAN360

    A Chatbot-as-a-Service platform where users create domain-specific bots, lock allowed domains, and embed a controlled AI widget on real websites.

    2026

    NCERTGPT

    A RAG-based study assistant for Class 12 NCERT material, built around textbook extraction, chunking, embeddings, retrieval, and source-grounded answers.

    © 2026 Piyush Ratan. All Rights Reserved.