Please complete the CAPTCHA to continue
A concise factual answer block for searchers comparing this professional profile.
Jung-A Kim is listed as Senior Data Scientist at Thermo Fisher Scientific, based in Greater Seattle Area, United States. AeroLeads shows a matched LinkedIn profile for Jung-A Kim.
Jung-A Kim previously worked as Professional development at Career Break and Senior Data Scientist at State Farm. Jung-A Kim holds Master'S Degree, Statistics from San José State University.
This section adds company-level context without repeating Jung-A Kim's masked contact details.
Review company-level records connected to Jung-A Kim before choosing the right outreach path.
I am a curious data scientist who deeply cares about the work as to how it can benefit the clients. My curiosity leads to critical thinking and consideration of various factors related to issues. From my past work experience, I learned the importance of frequent proactive communications with domain experts to solve the problem not only quickly but also in the right way.I developed technical skills in NLP, Computer Vision, and AI with exposure to try different APIs. The past projects include both classical and deep learning models for cases such as predicting time lags, injury severity, image classification, text classification using proper python packages. Through these experiences, I learned to drive effective communication with business partners with the help of proper tools related to the data labels understanding the crux of the problem early in the process.My academic background in MS Statistics built the foundation to achieve the necessary skills to land on the first role as a data scientist. The capstone project at school was a benchmark project of a new algorithm developed by Intuit data scientists. I presented the performance with simulated dataset along with the training progress. In the first year's summer, I took machine learning course at Stanford which had very meticulous curriculum developed by Andrew Ng entailing the anatomy of well-known ML algorithms. After graduation, I completed a data science bootcamp with four projects from brainstorming to implementation with e-commerce, Amazon review dataset, etc. The methods included sentiment analysis, text summarization, topic modeling, etc and Flask application for final demo.
Company context helps verify the profile and gives searchers a useful next step.
A career timeline built from the work history available for this profile.
Waltham, Ma, Us
Bringing LLM to enhance clinical trial processDatabricks AutoML for prototyping models for predictive analyticsSelf-driven analysis and presentation of outcome with business partnersTools: Databricks, Snowflake, AzureDevOps, AWS S3
Self-development for the next chapterAttending AI workshopsResearch domain knowledge
Bloomington, Illinois, Us
Generative AI prompt engineering for summary and extraction - Mistral AI 7B Instruct 2.0 to extract and parse JSON output into auto-population applicable format for prevention of multi-million dollar loss of missing due date of payment and important key information from an attorney. - Active communication with Claims partners to refine prompt with domain knowledge - Numerous prompt engineering trials for improvements in the response - Chunk responses for memory
Bloomington, Illinois, Us
Worked for AI solutions in Auto Injury & Vision & NLP to Enhance prioritization for claim handling process. • Computer Vision model deployed with F1 average above 92% on 11 categories of fire claims categories which is the highest performance of vision model developed so far considering the noise in image data from customer/agent uploads. Significantly shortened labeling time and cost with the millions of data in Auto claims. • Document Classifier with a transformer-base discriminator google ELECTRA model for multiclass classification. • Data Validation: voxel51 Fiftyone app to validate the labeled data with business partners. • AWS Training job: Explored and leveraged AWS Training Job to optimize instance usage time for training multiple models in parallel. Engaged in communication with MLE, training job in script mode successfully ran transformer models efficiently both time and cost-wise. • AutoGluon: Baseline modeling using AutoMM from AutoGluon. • PyTorch/PyTorch Lightning/Huggingface: For training large models, explored and developed distributed data processing in these three different APIs. Tensorflow Logger were used to monitor the distributed training in all cases. ResNet50, Densenet161, and Swin Transformers were mainly developed and compared. • XGBoost for predicting time lags between claim processes to expedite claim settlements which saves a lot of cost from delay. Bayesian Optimization, faster hyperparameter tuning with performance improvement. Object oriented programming in Python. • Injury Severity scale prediction made with injury text using BertModel. • Proactive communication with business partners during error analysis identifying labeling inconsistency. Visual aids of matplotlib for exploratory data analysis convinced the model performance.
New York, Ny, Us
Designed and built practical business solutions for e-commerce platforms in a 12-week immersive data science bootcamp.REVIEW-BASED SEARCH ENGINEDeveloped a search engine for e-commerce platforms that match between a query's keywords and reviews using sentiment scores and similarity scores.Created a user-interactive Flask app that retrieves optimal products with summarized reviewsTools: Universal Sentence Encoding(USE) for sentiment analysis, PEGASUS for summarization, Spacy and NLTK for preprocessingNETFLIX APP REVIEW TOPIC MODELINGMotivated to help developers to promptly deal with trouble-shooting and adjust their quality assurance.Trained BERT model to classify the reviews with tuned LDA parameters.Created an app for users to write a review and be referred to the right service. Wrote a blog about the difference between the bag-of-words scheme and context-base scheme. Published in Chatbot's Life Magazine.IEEE-CIS FRAUD DETECTION ON KAGGLEUsing online-transaction dataset on Kaggle, compared Bagging methods with respect to recall and precision. Achieved 0.87 AUC with XGBoost and Sigmoid Calibration on Kaggle submission. Permutation importance with recall ratio was used for interpreting eachfeature's contribution.Tools: xgboost, eli5, sklearn.calibration, TableauKEYBOARD SALE RATE PREDICTION BY POISSON REGRESSIONPredicted a keyboard's sale rate using Poisson regression. Handled feature engineering, imputation, and cross-validation.Web-scraped eBay query results using BeautifulSoup and improved R^2 and MAE by feature engineering and multiple imputation with MICE algorithm. Selected by Medium curators to be recommended to readers in ‘Data Science’ section.
Mountain View, California, Us
As a culminating experience for MS Statistics program, participated in a project to prove the performance of Finite Rank Deep Kernel Learning invented by the data scientists at Intuit.• Trained Finite Rank Deep Kernel Learning in PyTorch and compared its Normalized Root Mean Squared Error, GP inference time, and running time with Gaussian Process, Deep Kernel Learning, and Eigenvector decomposition kernel (Dataset: Boston Housing price in 1978).• Presented the main findings to the data scientists from industries and mathematics & statistics department.
San Jose, Ca, Us
• Gave students brief lectures about mathematical concepts and problems in workshop sessions for Calculus for Business/Aviation• Taught students with different approaches to the problems to help understanding and developing mathematical perspectives • Encouraged group discussions and freely asking questions
San Francisco, California, Us
• Developed a database administration tool in node.js for PostgreSQL functions using JavaScript, jQuery, and HTML5/CSS3.• Wrote various PostgreSQL DDL commands with nested queries and complex joins to retrieve database objects in the object browser and created a professionally designed interface for CRUD, index, view functions.• Researched broad graph database market and presented its competency for promoting the company's product.
Quick answers generated from the profile data available on this page.
Jung-A Kim works for Thermo Fisher Scientific.
Jung-A Kim is listed as Senior Data Scientist at Thermo Fisher Scientific.
Jung-A Kim is based in Greater Seattle Area, United States while working with Thermo Fisher Scientific.
Jung-A Kim has worked for Thermo Fisher Scientific, Career Break, State Farm, Metis, and Intuit.
You can use AeroLeads to view verified contact signals for Jung-A Kim at Thermo Fisher Scientific, including work email, phone, and LinkedIn data when available.
Jung-A Kim holds Master'S Degree, Statistics from San José State University.
Search by job title, company, industry, location, and seniority. Export verified B2B contact data when you need it.
Start free trial Search contacts