Hwanmun Kim Email & Phone Number

Mountain View, Ca, Us

Mountain View, Ca, Us

Campbell, Ca, Us

1. Developed LLM-based NL-to-SQL system- Developed a NL-to-SQL system based on few-shot prompts composed of systematically sampled examples- Achieved performance comparable to leading fine-tuning based systems in CoSQL benchmark using GPT3.5-turbo- Wrote and submitted a research paper to ACL conferences- Adapted developed system to SQL queries on clinical DB for potential application in products2. Developed LLM-based natural language inference (NLI) system for clinical trials- Designed a new NLI system composed of multiple modules using both of few-shot prompting and fine-tuned mistral model- Developed training and evaluation pipelines for the devised system- Trained the model and submitted the system to SemEval 2024 Task 2 challenge- Wrote a system paper for the challenge submission and got accepted for the corresponding ACL workshop3. Developed LLM-based lay summarization system for biomedical articles- Developed training and evaluation pipelines to experiment different fine-tuning approaches to develop summarization system targeting biomedical articles- Trained lay summarization models and submitted the models to BioLaySumm 2024 challenge, achieved the 3rd place overall in the final leaderboard- Wrote and submitted system paper for the challenge and got accepted for the corresponding ACL workshop4. Improving embedding model for biomedical documents for application in products- Generated synthetic data to train the embedding model using open-sourced biomedical documents and LLM-based prompting- Performed a case study to compare different querying methods using vector DB- Demonstrated proof-of-concept experiment of the latest embedding dimension reduction method

Santa Clara, California, Us

1. Developed speech processing models using ML techniques: - Developed language classification models & emotion recognition models. - Examined and cleaned datasets & set relevant evaluation metrics. - Implemented PyTorch classification models with LSTM, CNN, and self-attention architectures. - Implemented pipelines for data preprocessing, training, and evaluation. - Trained models and provided a prototype for the business development team.2. Improved Korean automatic speech recognition (ASR) models: - Trained Korean ASR models to improve the models used in company products. - Delivered a CTC-trained acoustic model for the business team.3. Improved voice morphing system: - Implemented evaluation pipeline and evaluated the performance of a pre-trained baseline model. - Replaced speaker embedding model with a model pre-trained for speaker verification.

College Park, Md, Us

1. Led multiple physics research projects in the cross-institutional team setting.2. Published research papers in top-tier journals and presented in conferences.3. Proposed experiments and validated the proposals with scientific computations and simulations: - Extensively used MATLAB, Python, C++ for the computations and simulations. - Visualized experiment schemes and computation results with MATLAB, Mathematica, Python. - Run computations in parallel on the high-performing computing cluster.

College Park, Md, Us

- Worked for Aug 2011-May 2012, Aug 2014-Dec 2014

Santa Clara, California, Us

- Mandatory military service in South Korea.- Worked as a meteorological observer.

Doctor Of Philosophy - Phd, Physics

Bachelor Of Science - Bs, Physics