Ali Abbas Baygi

About

I’m a Data scientist at Smartblick where I analyze data, design algorithms and develop machine learning softwares. I received my master degree in PEESE from TU Berlin. Throughout my study, I developed my knowledge in statistics, data science and machine learning. I would love to create small and real-world applications using machine learning models and sharing them through open source. I am also interested in using these engineering tools to understand and model industrial fields. I immigrated from Iran, where I studied mechanical engineering.

Research Interest

Deep Learning (Keras, TensorFlow, PyTorch)
Transformers for NLP (e.g. NER, Question Answering, PoS, Topic Modeling,..)
Large Language Models (Opneai API, LangChain for LLM Application Development)
Industrial internet of things (IIoT)
Sensor data analysis, Data recording and Data processing
Time series classification, segmentation, pattern recognition and Feature extraction
Deployment of Machine Learning Models (using Streamlit, FastAPI, Docker)

Education

MSc in Process, Energy and Environmental Systems Engineering (PEESE) @ Technische Universität Berlin
MSc. in Mechanical Engineering @ Khajeh Nasir Toosi University of Technology, Iran
BSc. in Mechanical Engineering @ Mazandaran University, Iran

Licenses & certifications

BERT Question Answering

NLP, Question Answering, Transformers.

Elasticsearch - Deep Learning - Haystack - Huggingface - FastAPI

ML Model Deployment

ML Model Deployment using FastAPI, Streamlit, and Docker.

MLOps - Docker - FastAPI - Streamlit

Fine-tune LLMs with LoRA

Fine-tuning of the Flan-T5 LLM from Hugging Face using a Parameter Efficient Fine-Tuning (PEFT) technique known as LoRA and assess the outcomes through ROUGE score evaluation.

LLMs - Huggingface - Peft

Startup Name Generator with LangChain

Startup Name Generator using LangChain, OpenAI, and Streamlit.

LangChain - OpenAI - Streamlit