Data Scientist • Machine Learning Engineer
I design data-centric, explainable, and interactive AI systems.
If you only check a few repos, these show my “core themes” best:
- Graph-RAG-Engine | Explainable, data-grounded RAG using vector search + knowledge graph (systems-thinking approach).
https://github.com/AmirhosseinHonardoust/Graph-RAG-Engine - Forecast-Factory | Streamlit forecasting + scenario simulation for business-friendly decision support.
https://github.com/AmirhosseinHonardoust/Forecast-Factory - Synthetic-Data-Artist | “Microscope” for tabular synthetic data realism (copula vs VAE, distribution & structure checks).
https://github.com/AmirhosseinHonardoust/Synthetic-Data-Artist - Data-Storytelling-Dashboard | Narrative analytics dashboard (KPIs + cohorts + retention + takeaways).
https://github.com/AmirhosseinHonardoust/Data-Storytelling-Dashboard - Sentiment-Analysis-BERT | Clean transformer fine-tuning pipeline + evaluation artifacts.
https://github.com/AmirhosseinHonardoust/Sentiment-Analysis-BERT - Think-Like-a-Data-Scientist | Long-form mindset + framing + experimentation guide.
https://github.com/AmirhosseinHonardoust/Think-Like-a-Data-Scientist
- Build end-to-end ML: problem framing → data → modeling → evaluation → deployment (Streamlit / FastAPI).
- Focus areas:
- Synthetic data & robustness (fidelity, coverage, privacy leakage, downstream ML utility)
- Forecasting & decision tools (scenario simulation, interactive dashboards)
- LLMs + RAG + hybrid systems (traceability, grounding, graphs)
- Explainability & human-centered AI (SHAP, interpretability, “why” behind predictions)
Open to: interesting ML/DS roles and serious collaborations (product-y ML, data apps, applied research).
- Languages: Python, SQL, MQL5, Solidity
- ML/DL: PyTorch, TensorFlow/Keras, scikit-learn, XGBoost, LightGBM
- Time series: ARIMA/SARIMA, Prophet, LSTMs, rolling windows
- NLP: BERT, TF-IDF, classic ML (LogReg, Naive Bayes, SVM)
- Apps/Services: Streamlit, FastAPI, Plotly, SQLite/SQLAlchemy
- Explainability/analysis: SHAP, feature importance, evaluation frameworks, synthetic benchmarks
- Synthetic Data, Data Realism & Anomaly Detection
- Forecasting, Dashboards & Interactive ML
- LLMs, RAG & Hybrid AI Systems
- Core ML, Deep Learning & Portfolio
- Explainability & AI Philosophy
- Smart Contracts, Security & Risk Analytics
Generating realistic tables, probing data “authenticity”, and stress-testing models.
Top picks:
- Autocurator-Synthetic-Data-Benchmark | Synthetic tabular benchmarking (fidelity/coverage/privacy/utility + visual reports).
https://github.com/AmirhosseinHonardoust/Autocurator-Synthetic-Data-Benchmark - Synthetic-Data-Artist | Copula vs VAE deep dive + realism diagnostics.
https://github.com/AmirhosseinHonardoust/Synthetic-Data-Artist - Anomaly-Detection | End-to-end anomaly detection with injected realistic “weirdness”.
https://github.com/AmirhosseinHonardoust/Anomaly-Detection
More in this cluster
- Market-Basket-Analysis | Apriori/FP-Growth + association rules
https://github.com/AmirhosseinHonardoust/Market-Basket-Analysis - Sales-Data-Analysis | Synthetic sales generation + KPIs + trends
https://github.com/AmirhosseinHonardoust/Sales-Data-Analysis - Missing-Data-Doctor | Missingness profiling + imputation + before/after ML impact
https://github.com/AmirhosseinHonardoust/Missing-Data-Doctor - Noise-Injection-Techniques | Robustness experiments via controlled noise
https://github.com/AmirhosseinHonardoust/Noise-Injection-Techniques
Treat forecasting and analytics as interactive tools, not static reports.
Top picks:
- Forecast-Factory | Upload series → forecast + confidence intervals → “what-if” driver simulations.
https://github.com/AmirhosseinHonardoust/Forecast-Factory - Market-IQ | KPI-focused BI-style analytics app for transaction data.
https://github.com/AmirhosseinHonardoust/Market-IQ - Data-Storytelling-Dashboard | Cohorts, retention curves, narrative takeaways.
https://github.com/AmirhosseinHonardoust/Data-Storytelling-Dashboard
More in this cluster
- Beyond-Charts-Interactive-Storytelling | Essay + interactive views (RFM, cohorts, lifecycle)
https://github.com/AmirhosseinHonardoust/Beyond-Charts-Interactive-Storytelling - AI-Report-Factory | Repeatable report generation (KPIs + charts + narrative)
https://github.com/AmirhosseinHonardoust/AI-Report-Factory - AI-Personal-Study-Tracker | Streamlit + SQLite + ML to surface patterns
https://github.com/AmirhosseinHonardoust/AI-Personal-Study-Tracker - Demand-Forecasting | ARIMA/SARIMA template
https://github.com/AmirhosseinHonardoust/Demand-Forecasting - ML-Playground-Autodetect | Upload dataset → auto detect task → baseline pipeline + eval
https://github.com/AmirhosseinHonardoust/ML-Playground-Autodetect
Explainable, data-grounded LLM systems with retrieval & graphs.
Top picks:
- Graph-RAG-Engine | FAISS retrieval + knowledge graph + traceability (FastAPI + optional Streamlit).
https://github.com/AmirhosseinHonardoust/Graph-RAG-Engine - Designing-Hybrid-AI-Systems | Patterns, failure modes, interpretability for hybrid intelligence.
https://github.com/AmirhosseinHonardoust/Designing-Hybrid-AI-Systems - RAG-vs-Fine-Tuning | Practical decision framework (cost/latency/maintenance/data constraints).
https://github.com/AmirhosseinHonardoust/RAG-vs-Fine-Tuning
Classic ML projects with clean structure and clear evaluation.
Top picks:
- Stock-LSTM-Forecasting | PyTorch LSTM + sliding windows + plots
https://github.com/AmirhosseinHonardoust/Stock-LSTM-Forecasting - Image-Captioning-CNN-LSTM | ResNet encoder + LSTM decoder + BLEU + examples
https://github.com/AmirhosseinHonardoust/Image-Captioning-CNN-LSTM - Sentiment-Analysis-BERT | Fine-tune + confusion matrix + ROC/PR curves
https://github.com/AmirhosseinHonardoust/Sentiment-Analysis-BERT
More in this cluster
- Sentiment-Analysis-NLP | TF-IDF + classical ML
https://github.com/AmirhosseinHonardoust/Sentiment-Analysis-NLP - Movie-Recommendation-System | Content + collaborative filtering (matrix factorization)
https://github.com/AmirhosseinHonardoust/Movie-Recommendation-System - LSTM-Time-Series-Forecasting | General LSTM forecasting template
https://github.com/AmirhosseinHonardoust/LSTM-Time-Series-Forecasting - Handwritten-Digit-GAN | DCGAN on MNIST + latent interpolation
https://github.com/AmirhosseinHonardoust/Handwritten-Digit-GAN
Code + essays about how we think about and interpret AI systems.
Top picks:
- Shap-Mini | Tiny SHAP demo that’s easy to learn from
https://github.com/AmirhosseinHonardoust/Shap-Mini - Think-Like-a-Data-Scientist | Framing → hypotheses → experiments → insight → action
https://github.com/AmirhosseinHonardoust/Think-Like-a-Data-Scientist - The-Future-of-Interactive-ML | Why UI/UX changes modeling choices
https://github.com/AmirhosseinHonardoust/The-Future-of-Interactive-ML
More in this cluster
- Forecasting-The-Future-of-Forecasting
https://github.com/AmirhosseinHonardoust/Forecasting-The-Future-of-Forecasting - Algorithmic-Empath-Human-Fallibility
https://github.com/AmirhosseinHonardoust/Algorithmic-Empath-Human-Fallibility - Measuring-The-Soul-of-Data
https://github.com/AmirhosseinHonardoust/Measuring-The-Soul-of-Data - Quiet-Machines-Minimalist-AI
https://github.com/AmirhosseinHonardoust/Quiet-Machines-Minimalist-AI
Applying ML & static analysis ideas to smart contracts and risk.
- Python-Solidity-Feature-Engineering | Parse Solidity + extract structural/semantic features
https://github.com/AmirhosseinHonardoust/Python-Solidity-Feature-Engineering - Smart-Contract-Risk-Analyzer | Heuristics + static analysis → risk scoring / triage direction
https://github.com/AmirhosseinHonardoust/Smart-Contract-Risk-Analyzer
“AI is not just about models; it’s about systems that solve real problems for real people.”