Tae-Gil Noh Email & Phone Number

Berlin, Germany

- Develping and applying machine learning methods on customer support automation. Responsible for the company’s back-end data processing engine, which processes customer support data of diverse European B2C companies. - Developed 3 generations of NLU (natural language understanding) engines, and launched them into productive service. Word-vector based processing suit at 2018, BERT-based suits at 2020, and LLM (GPT-3.5 and GPT-4) based suits at 2023. - Gathered vast experience (probably more than I wanted to) on the issues of using ML models in real world applications: this includes not only about training data collection and model trainings, but also on the operational issues such as deployments, model versioning & maintenance, optmization of inference speed on CPU/GPU instances, etc.

Heidelberg

As a data scientist in Smart Energy Group. - Built prediction and optimization models for various European / Japanese energy domain data.- Developed grid control logics by combining predictive analysis and contrained optimization methods, as a part of EU FP7 OrPHEuS project. Led the main work package of the project (control logic development and executing experiments) as the work package leader.

Heidelberg

Postdoc years. Applied various machine learning techniques on Natural Language Processing. - Developed a multilingual open source platform for Textual Inference, as a part of EU FP7 project EXCITEMENT. Designed and wrote the platform architecture, then served as a member of development team and provided one of the main inference methods. - Developed and proposed Conditioned Language Model (CLM): a meta-framework for language model (LM), which enables better utilization of contextual information via dynamic weighting of corpus data.

Buk-Gu, Daegu, Korea

- PhD topic: graph kernels - with applications of representing entities that are naturally represented in graph strauctures, including web users and word usages. - Developed new methods of natural language (NL) query processing that translate NL queries into formal queries like SPARQL. - Designed and implemented a semantic search system with the goal of finding “emergent” (statistically emerging) knowledge from social web contents.

Buk-Gu, Daegu, Korea

- Led a team of engineers in developing an XHTML/HTML rendering engine for a mobile Web browser, with a goal of small memory footage and good software maintainability. - Participated in porting and customization of mobile software suites including messaging modules and browsers for Samsung mobile devices.

Doctor Of Philosophy (Phd), Computer Engineering

Machine Learning Lab, under the supervision of Prof. Seong-Bae Park.

Master'S Degree, Computer Engineering

Bachelor (1999) and Master (2001) for Computer Engineering