A passionate ML & NLP engineer with expertise in web development and a focus on under-resourced languages.
I am a Machine Learning and Natural Language Processing enthusiast with a strong background in web development. I completed my Master's in Computer Science (Informatics) at the Technical University of Munich, focusing on Deep Learning and NLP.
My research interests include language modeling for under-resourced languages, personality trait estimation, and applying machine learning to solve real-world problems. I have experience working with various machine learning frameworks. Furthermore, I have published papers in international conferences.
Currently based in Freising, Germany, I am passionate about developing innovative solutions that bridge the gap between technology and linguistics.
My technical toolkit and areas of expertise
My academic and professional journey
Aug 2022 - Present
Koinon - TU Munich
Apr 2024 - Dec 2024
Schütze Lab, Institute for Information and Language Processing, LMU
Sep 2023 - Dec 2023
Technical University of Munich – School of Management
Oct 2021 - Dec 2024
Technical University of Munich
Oct 2017 - Oct 2018
CSE Department, Ahsanullah University of Science and Technology
Oct 2014 - Dec 2018
Ahsanullah University of Science and Technology
A selection of my research and development work
Developed an automatic pipeline to crawl the web for links in under-resourced languages and built a textual dataset for language modeling and perplexity analysis.
A Python application that leverages multimodal LLM to identify Pokémon from webcam images and retrieve detailed information about them.
Developed a prototype RAG chatbot leveragin OpenAI for university administration that provides user support based on question regarding user portal.
Evaluated zero-shot performance of GPT 3.5 and 4 for legal sentence labeling using OpenAI's API.
Developed a crawling agent leveraging Selenium and Beautiful Soup to curate social media texts. Scraped and analyzed social media content of 6,000 startup founders to estimate their OCEAN personality traits and explored the relationship with startup success.
Compared various deep learning approaches (BiLSTM-CRF, T5, Seq2Seq, LegalBERT) for rhetorical role labeling in Indian legal documents.
Reprojected NDVI images from NASA's satellites to estimate rice yield using ML algorithms. Published at IEEE IGARSS 2019, Yokohama, Japan.
My research contributions to the scientific community
DOI: 10.1109/IGARSS.2019.8899084
Published at IEEE IGARSS 2019, Yokohama, Japan
Citations: 15
DOI: 10.1109/SMART46866.2019.9117468
Citations: 9
DOI: 10.18034/ra.v8i1.449
Citations: 3
Let's connect and discuss potential collaborations
Email: in.sefat@tum.de
Phone: +49 157 5289 4251
Location: Freising, Germany