Amir

Fake-News-Detector. A professional TF-IDF + Logistic Regression style-risk classifier for educational fake-news detection, with a Streamlit dashboard, honest evaluation, uncertainty handling, and leakage analysis.

Image-Captioning-CNN-LSTM. An end-to-end image captioning project using a CNN encoder (ResNet-50) and LSTM decoder in PyTorch. Includes vocabulary building, preprocessing, training with BLEU evaluation, and inference. Generates natural language captions for images with saved metrics, model checkpoints, and visualization outputs.

Coffee-Shop-Profit-Predictor. Predict the profitability of potential coffee shop locations using SQL and Python. Combines data engineering with feature-rich regression modeling, visual analytics, and business insights to support data-driven site selection and retail decision-making.

Stock-LSTM-Forecasting. Predict stock prices using LSTM networks in PyTorch. This project covers data preprocessing, sliding window creation, model training with early stopping, and evaluation with RMSE/MAE/MAPE. Includes visualizations of training loss, predicted vs actual prices, and short-horizon forecasts.

Sentiment-Analysis-BERT. End-to-end sentiment analysis of tweets using BERT. Includes preprocessing, training, and evaluation with classification reports, confusion matrices, ROC curves, and word clouds. Demonstrates fine-tuning of transformer models for text classification with modular, reproducible code.

AmirhosseinHonardoust. I’m Amirhosein Honardoust, a passionate developer and data scientist with experience in Python, PyTorch, deep learning, and machine learning projects across vision, NLP, and forecasting. I enjoy turning data into insights and building portfolio-ready AI applications that combine practicality with creativity.

Market-Basket-Analysis. Python project for Market Basket Analysis. Generates synthetic retail transactions, mines frequent itemsets using Apriori & FP-Growth, derives association rules, and outputs CSVs + visualizations. Portfolio-ready example demonstrating data science methods for uncovering product co-purchase patterns.

AI-Productivity-Tracker. Analyze and predict daily productivity using SQL, machine learning, and psychology. This project combines behavioral data, circadian rhythm analysis, and ElasticNet regression to model focus, stress, and performance, transforming work patterns into actionable insights.

Employee-Performance-Analytics. Analyze employee performance and productivity using SQL and Python. Build HR dashboards to evaluate departmental KPIs, efficiency trends, and workload balance. Generates actionable insights through data visualization and KPI aggregation for business decision-making.

Demand-Forecasting. End-to-end demand forecasting with Python using synthetic time-series sales data. Includes data generation, cleaning, ARIMA/SARIMA model selection by AIC, evaluation with RMSE and MAPE, and 90-day forecasts with confidence intervals. Reproducible scripts and visualizations for portfolio showcase.

AI-Personal-Study-Tracker. An AI-driven productivity tracking app built with Python, Streamlit, SQLite, and Machine Learning. It logs and analyzes study sessions, predicts productivity using Random Forest models, and visualizes key insights to help learners improve focus, habits, and overall academic efficiency.

Movie-Recommendation-System. Movie recommendation system with Python. Implements content-based filtering (TF-IDF + cosine similarity), collaborative filtering with matrix factorization (TruncatedSVD), and a hybrid approach. Evaluates with Precision@K, Recall@K, and NDCG. Includes rating distribution plots, top movies, and sample recommendations.

Fraud-Detection-SQL-Supervised. Detect and classify fraudulent transactions using SQL and Python. Generate behavioral features with SQLite, train a Logistic Regression model, and evaluate performance with AUC, precision, recall, and ROC analysis. A complete supervised fraud detection workflow.

Sales-Data-Analysis. Synthetic sales data analysis with Python. Generate realistic sales transactions, clean and validate data, compute KPIs, and visualize revenue trends by day, month, and category. Includes reproducible scripts and charts for portfolio demonstration.

Handwritten-Digit-GAN. A PyTorch implementation of a Deep Convolutional GAN (DCGAN) trained on MNIST. Includes training scripts, generator & discriminator models, random sample generation, latent space interpolation, and loss curve visualization to create realistic handwritten digit images.

Algorithmic-Empath-Human-Fallibility. A deep exploration of Algorithmic Empathy, the next frontier in AI understanding. This project examines how machines can learn from human fallibility, model disagreement, and align with moral reasoning. It blends psychology, fairness metrics, interpretability, and co-learning design into one framework for humane intelligence.

LSTM-Time-Series-Forecasting. A hands-on project for forecasting time-series with PyTorch LSTMs. It creates realistic daily data (trend, seasonality, events, noise), prepares it with sliding windows, and trains an LSTM to make multi-step predictions. The project tracks errors with RMSE, MAE, MAPE and shows clear plots of training progress and forecast results.

Sales-Insights-SQL. Analyze retail sales data using SQL and Python. Build a SQLite database from CSV, run SQL queries for key KPIs (revenue, top products, AOV, trends), and visualize results with Matplotlib. A portfolio-ready project demonstrating SQL + data analytics + reporting automation.

Beyond-Charts-Interactive-Storytelling. A comprehensive guide and codebase for building interactive storytelling dashboards with Python, Streamlit, and Plotly. Learn how to transform static analytics into dynamic, user-driven data experiences that engage and inspire, featuring RFM segmentation, cohort analysis, and real-world insights.

Fake-Review-Detector. An AI-powered Fake Review Detector built with Python, Streamlit, and Scikit-learn. Uses TF-IDF vectorization, Logistic Regression, and behavioral text analytics (sentiment, exclamations, clichés) to identify synthetic or spammy product reviews. Includes training scripts and a full interactive dashboard.

Image-Classification-CNN. Image classification with PyTorch using Convolutional Neural Networks (CNNs). Trains on MNIST with convolution, pooling, and fully connected layers. Achieves over 99% accuracy with early stopping and checkpoints. Includes training/evaluation scripts, metrics, confusion matrix, training curves, and sample prediction visualizations.

Fraud-Detection-SQL-Unsupervised. Detect suspicious financial transactions using SQL and Python. Build user-level behavioral features in SQLite, apply Isolation Forest for anomaly detection, and visualize high-risk patterns. Demonstrates unsupervised fraud analytics and SQL-driven data science workflow.

Data-Storytelling-Dashboard. A fully interactive data storytelling dashboard for e-commerce analytics. Built with Python, Streamlit, and Plotly, it transforms transactional data into actionable insights through KPIs, cohort retention, RFM segmentation, and global visualizations, perfect for analysts and data scientists.

Sentiment-Analysis-NLP. Customer reviews sentiment analysis with Python and NLP. Generates a synthetic dataset of positive, neutral, and negative reviews, applies preprocessing (tokenization, stopwords, lemmatization), and builds TF-IDF features. Trains classifiers (Naive Bayes, Logistic Regression, Random Forest) with evaluation, confusion matrix and top features.

Customer-Churn-Prediction. Customer churn prediction with Python using synthetic datasets. Includes data generation, feature engineering, and training with Logistic Regression, Random Forest, and Gradient Boosting. Improved pipeline applies hyperparameter tuning and threshold optimization to boost recall. Outputs metrics, reports, and charts.

Anomaly-Detection. Anomaly detection in synthetic transaction and sales data with Python. Generates realistic data, injects unusual events, and applies Isolation Forest, Local Outlier Factor, and Z-score methods to detect outliers. Produces anomaly reports and visualizations for portfolio-ready demonstration of data science skills.

RAG-vs-Fine-Tuning. A comprehensive, professional guide explaining the differences, strengths, and best practices of Retrieval-Augmented Generation (RAG) and Fine-Tuning for LLMs, including workflows, comparisons, decision frameworks, and real-world hybrid AI use cases.

Graph-RAG-Engine. An explainable AI system that combines Graph Intelligence, Vector Search, and Retrieval-Augmented Generation (RAG) to deliver grounded answers and transparent reasoning paths. Includes a FastAPI backend, Streamlit UI, FAISS vector index, and an in-memory knowledge graph for hybrid retrieval and recommendations.

Community-Price-Tracker. A community-driven app for tracking and comparing the prices of everyday goods across cities. Built with Python, SQLite, Pandas, and Streamlit, it lets users log, visualize, and analyze item price trends, inflation, and cost-of-living differences with clear charts and full local data privacy.

Content-KPI-Monitor. A Streamlit + SQLite + Python dashboard for monitoring content performance KPIs. Tracks impressions, clicks, conversions, CTR, and conversion rates across categories and time. Includes an automated ETL from CSV → SQL view, interactive charts, and filters for data-driven content optimization.

TFIDF-vs-Word2Vec. A detailed educational guide explaining two essential NLP techniques, TF-IDF and Word2Vec. Learn how text is transformed into numerical vectors, compare their mathematical foundations, explore real-world use cases, and implement both methods in Python for text analysis and machine learning.

AI-Report-Factory. AI Report Factory automates data storytelling, transforming raw business data into analytical, narrative-driven reports. It ingests structured datasets, computes KPIs, generates visualizations, and produces Markdown & HTML reports with actionable insights for startups, analysts, and enterprises.

Forecast-Factory. Forecast factory is an interactive AI-powered forecasting and simulation tool built with Python, Streamlit, Prophet, and SQL. It enables analysts to forecast business metrics, run what-if scenarios, and visualize results in real time, transforming predictive analytics into actionable strategic simulations.

Think-Like-a-Data-Scientist. A thought-provoking article exploring how to think like a data scientist, without writing a single line of code. Covers the mental framework behind curiosity, structured reasoning, hypothesis testing, and data-driven storytelling, helping readers build analytical intuition beyond tools or syntax.

Forecasting-The-Future-of-Forecasting. A deep exploration of AI-driven foresight, how predictive models evolve into strategic collaborators. This repository presents Forecasting the Future of Forecasting, a professional essay on reflexive intelligence, human–AI collaboration, and the design of adaptive, explainable forecasting systems.

How-Streamlit-Makes-AI-Accessible. A detailed educational article exploring how Streamlit revolutionizes AI app development. Learn how this Python framework bridges the gap between data science and usability, empowering anyone to deploy interactive machine learning models without front-end coding or complex infrastructure.

The-Future-of-Interactive-ML. An in-depth exploration of the rise of human-centered, interactive machine learning. This article examines how Streamlit enables collaborative AI design by merging UX, visualization, and automation. Includes theory, architecture, and design insights from the ML Playground project.

Quiet-Machines-Minimalist-AI. A long-form essay exploring the philosophy of minimalist AI, how future intelligent systems can be calm, ethical, and invisible. Inspired by calm technology, design minimalism, and cognitive science, Quiet Machines envisions a world where the best technology listens more than it speaks.

Customer-Sentiment-Intelligence-Platform. An enterprise-grade NLP + Streamlit + SQL platform for analyzing customer feedback. Performs automated sentiment detection, stores labeled reviews in SQLite, and delivers real-time dashboards with probability insights to support business, marketing, and product optimization decisions.

The-Analysts-Mirror-Reflective-Dashboards. A professional exploration of reflective analytics, dashboards that evolve through user interaction. This project reimagines BI systems as adaptive, cognitive mirrors that learn analyst behavior, anticipate reasoning patterns, and transform data visualization into a collaborative, intelligent process.

Designing-Hybrid-AI-Systems. Hybrid AI is the future of explainable intelligence. This article explores how combining vector search, knowledge graphs, and retrieval-augmented generation (RAG) creates AI systems that can reason, cite, and explain their answers with insights learned from building a real Graph-Powered RAG Engine.

Synthetic-Data-Artist. A professional, research-grade comparison of Gaussian Copula and Variational Autoencoder (VAE) methods for synthetic tabular data generation. Includes full evaluation pipeline with distribution overlap, correlation analysis, PCA projections, pairplots, metrics, and automated visual reports.

Onchain-Security-Suite. A complete Web3 security toolkit combining AI-powered token auditing, ML-based deployer reputation scoring, and live Etherscan V2 data. Includes static analysis for rugpull detection, RandomForest reputation modeling, contract-fetching automation, and Solidity on-chain registries for transparent, reproducible security insights.

Smart-Contract-Risk-Analyzer. A lightweight static analysis engine for Solidity smart contracts. Extracts code features, detects dangerous patterns (delegatecall, tx.origin, call.value), computes heuristic risk scores, and classifies contracts into Low/Medium/High risk levels. Includes multiple example vulnerabilities and a clean CLI for rapid security assessment.

ML-Playground-Autodetect. A Streamlit-powered machine learning playground that automatically detects classification or regression tasks, builds pipelines with preprocessing, trains models interactively, and visualizes metrics using Plotly. Backward-compatible, fully responsive, and deployable on Streamlit Cloud or Docker.

Market-IQ. MarketIQ is a full-stack Streamlit + SQL + Prophet dashboard for real-time business intelligence. It transforms raw sales data into KPIs, trend analytics, and six-month forecasts. With SQL-driven insights, dynamic filtering, and exportable reports, MarketIQ helps analysts visualize growth and predict performance instantly.

Teaching-Neural-Networks-to-Imagine-Tables. A comprehensive deep dive into how Variational Autoencoders (VAEs) learn to generate realistic synthetic tabular data. This project explores latent space learning, probabilistic modeling, and neural creativity, combining data privacy, interpretability, and generative AI techniques in a structured format.

PR-Guardian-AI. PR Guardian AI automatically reviews pull requests using advanced AI analysis. It identifies code issues, security risks, performance problems, and provides actionable suggestions directly inside GitHub pull requests.

Noise-Injection-Techniques. Noise Injection Techniques provides a comprehensive exploration of methods to make machine learning models more robust to real-world bad data. This repository explains and demonstrates Gaussian noise, dropout, mixup, masking, adversarial noise, and label smoothing, with intuitive explanations, theory, and practical code examples.

Cognitivelens-AI-Human-Comparison. CognitiveLens is a Streamlit-powered analytics tool for exploring alignment between human and AI decisions. It visualizes fairness, calibration, and interpretability through metrics like Cohen’s κ, AUC, and Brier score. Designed for ethical AI, bias auditing, and decision transparency in machine learning systems.

Shap-Mini. A minimal, reproducible explainable-AI demo using SHAP values on tabular data. Trains RandomForest or LogisticRegression models, computes global and local feature importances, and visualizes results through summary and dependence plots, all in under 100 lines of Python.

How-AI-Detects-Rugpulls. A deep technical article exploring how AI, feature engineering, and static smart-contract analysis uncover rugpull risks before humans detect them. Covers Solidity pattern mining, mint abuse detection, blacklist/fee manipulation signals, ML-inspired scoring models, and how to quantify ERC-20 token scam probability.

Autocurator-Synthetic-Data-Benchmark. Autocurator is a comprehensive benchmarking toolkit for evaluating synthetic tabular data. It measures fidelity, coverage, privacy, and utility through quantitative metrics, visual reports, and PCA/correlation diagnostics. Ideal for validating VAE, GAN, Copula, or Diffusion-generated datasets.

The-Illusion-of-Sudden-Failure. A systems-thinking essay that explains why failure rarely happens suddenly. It shows how slow drift, accumulating pressure, and weakening buffers push systems toward collapse long before outcomes change, and why prediction-focused analytics miss the most important phase of failure.

Measuring-The-Soul-of-Data. A narrative and technical exploration of data authenticity through the four pillars of synthetic data realism, Fidelity, Coverage, Privacy, and Utility. This thought-leadership piece combines storytelling, mathematics, and code to explain how these metrics define the ethical and functional “soul” of data in AI systems.

ML-Powered-Token-Launch-Auditor. A hybrid Solidity + Python security toolkit that analyzes ERC-20 token contracts using static pattern extraction and ML-inspired scoring. Detects mint backdoors, blacklist controls, fee manipulation, trading locks, and rugpull mechanics. Outputs interpretable risk scores, labels, and structured features for deeper analysis.

ML-Meets-Etherscan-Realtime-Scanner. AI-powered real-time smart contract scanner that connects Machine Learning with Etherscan V2 to analyze newly deployed contracts instantly. Fetches verified Solidity code, performs static risk analysis, computes ML-driven deployer trust scores, and generates full security intelligence pipelines for Web3 threat detection.

Python-Solidity-Feature-Engineering. A practical, research-friendly toolkit demonstrating how Python can read, parse, and analyze Solidity smart contracts using feature-engineering techniques. Extracts structural and security-relevant signals from Solidity code, detects risky patterns, builds interpretable features, and forms the basis for heuristic or ML-driven security analysis.

Machine-Learning-Warning-Systems. A long-form article and practical framework for designing machine learning systems that warn instead of decide. Covers regimes vs decimals, levers over labels, reversible alerts, anti-coercion UI patterns, auditability, and the “Warning Card” template, so ML preserves human agency while staying useful under uncertainty.

The-Twin-Test-High-Stakes-ML. A long-form article introducing the Twin Test: a practical standard for high-stakes machine learning where models must show nearest “twin” examples, neighborhood tightness, mixed-vs-homogeneous evidence, and “no reliable twins” abstention. Argues similarity and evidence packets beat probability scores for trust and safety.