Hackerz Institute
of Tech Innovation
Enroll

Lab Projects

Real-world applications built by our students and team — showcasing the practical skills learned at Hackerz Institute.

AI / NLP Internal AI study tool

Synapse Agent

An adaptive AI study companion built with Next.js and Groq, featuring Socratic tutoring, bilingual interaction, and voice mode.

Next.js 15 Groq API Llama 3.1
Repo
AI / EdTech Student learning platform

EduSathi Platform

A role-based AI learning platform using RAG and LLMs to transform student PDFs into AI tutoring, adaptive quizzes, and flashcards.

Python LangChain FAISS
Repo
Security / ML Security Operations

Mini-SIEM AI

An AI-powered mini-SIEM combining Random Forest attack classification, SHAP explainability, and LLM-based threat reasoning.

Scikit-learn SHAP Streamlit
Repo
Security / AI Live project

Phishing Detector Advanced

A multilingual five-layer phishing detection platform combining signal-based ML, URL intelligence, HTML email analysis, and SMTP checks.

Python Regex DNS Analysis Llama 3.1
Repo

Synapse Agent

Synapse is an ultra-fast, adaptive AI study companion designed to make learning more interactive and less overwhelming. It uses conversational AI, Socratic learning, micro-learning responses, bilingual support, voice interaction, and local session persistence.

Key Features

  • Socratic-style AI tutoring instead of simply giving answers
  • Concise responses designed to reduce cognitive overload
  • English and conversational Hindi support
  • Ultra-low-latency streaming responses
  • Voice mode using the Web Speech API
  • Rolling conversation context & Local study-session history

Technology

Next.js 15, React 19, CSS3, Groq API, Llama 3.1 8B Instant, Web Speech API, Local JSON storage, async-mutex.

View on GitHub

EduSathi Platform

EduSathi is a production-oriented AI educational platform that allows students to learn directly from their own study materials. It uses Retrieval-Augmented Generation (RAG), LLMs, adaptive quizzes, exam simulations, flashcards, analytics, and role-based access control.

Student Features

  • AI Chat Tutor grounded in uploaded PDFs
  • Adaptive AI-generated MCQs & Timed exam simulation
  • Automatic flashcard generation & Topic mastery tracking

RAG Pipeline

PDF → Text Extraction → Chunking → Sentence Transformers → FAISS → Relevant Context → LLaMA 3.3 70B → Grounded Response

Technology

Python 3.10+, Streamlit, LangChain, FAISS, Sentence Transformers, Groq API, LLaMA 3.3 70B, SQLite, bcrypt, pdfplumber, PyPDF2.

View on GitHub

Mini-SIEM AI

Mini-SIEM is a security analytics platform that combines machine-learning attack classification, explainable AI, risk scoring, LLM threat reasoning, automated incident reports, and an interactive security dashboard.

Detection Pipeline

Raw Network Logs → Feature Engineering → Random Forest → Attack Classification → Risk Scoring + SHAP → LLM Threat Reasoning → Incident Report → Dashboard

Key Features

  • Network feature engineering (Nine engineered ML features)
  • Random Forest classification & GridSearchCV hyperparameter tuning
  • SHAP TreeExplainer & 0–100 composite risk scoring
  • GPT-4o-mini threat reasoning & Automated incident-response PDF generation
  • Interactive Streamlit SOC-style dashboard

Reported Performance

Weighted F1: 0.978. Class-level precision/recall/F1 around 0.94–0.99 on the sample dataset.

View on GitHub

Phishing Detector Advanced

Phishing Detector Advanced is a five-layer phishing detection system designed to identify phishing attempts using signal-based machine learning, URL intelligence, email analysis, SMTP validation, and AI reasoning. It specifically supports English, Hindi, and Hinglish to address common phishing and fraud patterns in the Indian threat landscape.

Detection Layers

  • Text Analysis (25%): Keywords, urgency, threats, English/Hindi/Hinglish
  • URL Analysis (30%): TLDs, DNS, lookalike domains, IP URLs
  • HTML Email Analysis (25%): Forms, hidden elements, link mismatches
  • SMTP Analysis (20%): DKIM, SPF, DMARC, From/Reply-To mismatches
  • AI Assessment: LLM-based threat explanation

Reported Performance

Accuracy: 97% | Precision: 96% | Recall: 97% | F1 Score: 97%

View on GitHub
Hackerz Institute
of Tech Innovation
Story Blog Courses Lab Projects Reach Out
Start Building Now