I work where machine-learning research meets real systems, with a focus on LLM/RAG systems, information retrieval, and adaptive time-series forecasting.
- 🔬 Building reproducible experiments, data/model pipelines, and careful evaluation workflows.
- 🧠 Exploring semantic retrieval, embeddings, and retrieval-augmented generation.
- 📈 Developing adaptive approaches to financial time-series forecasting.
- 👁️ Experimenting with deep learning, computer vision, speech, and sequence modeling.
- 🎓 Interested in research collaborations and PhD opportunities in AI and Machine Learning.
Clear questions. Reproducible experiments. Honest evaluation. Useful systems.
| Project | Focus |
|---|---|
| AIR-CEEMDAN Forecasting · manuscript submitted | Adaptive CEEMDAN-based forecasting with hierarchical LSTM and Ridge models, rolling out-of-sample evaluation, ablations, and statistical comparison. |
| FinQuery | A banking QA bot using fine-tuned BGE embeddings, FAISS retrieval, Kafka, Telegram, and Dockerized services. |
| LLM & Retrieval Experiments | Embedding fine-tuning, FAISS, Chroma, LangChain, and RAG prototypes. |
| Persian Semantic Search | Persian preprocessing, inverted indexing, TF-IDF retrieval, and Word2Vec-based semantic expansion. |
Explore more projects
| Project | Focus |
|---|---|
| Financial Time-Series Forecasting | Financial forecasting experiments with LSTM/BiLSTM models and multi-metric evaluation. |
| Adaptive Trading System Design | Architecture study combining regime analysis, structural patterns, volatility/volume signals, and sequence models. |
| NLP Language Modeling | Statistical n-grams and neural language models using RNN, LSTM, GRU, and BiLSTM. |
| Facial Emotion Recognition | FER-2013 experiments with classical baselines, CNNs, and OpenCV-based face detection. |
| Spoken Digit Recognition | MFCC audio features with neural and recurrent classification models. |
| Airflow Supermarket Analytics | A retail data-cleaning, storage, and analytics pipeline orchestrated with Apache Airflow. |
| Distributed Data Processing Labs | Spark, MapReduce, MRJob, MongoDB, and Python multiprocessing experiments. |
| Machine Learning Foundations | From-scratch neural networks and foundational machine-learning experiments. |
Also in development: an LLM-assisted professor–project matching pipeline for evidence-based academic recommendation.
Interested in ML research, LLM systems, information retrieval, or adaptive forecasting? Let's talk about research collaborations and PhD opportunities.
linkedin.com/in/moslem-amini-020932223
Research with rigor. Build with purpose. Share what can be reproduced.