Capture Questions, Answer with Code

• Name: Young Jin Ahn
• Email: [email protected]
• Blog: https://snoop2head.github.io/

Host / Platform	Topic / Task	Result	Repository	Year
National IT Industry Promotion Agency	Machine Reading Compehension	🥈 2nd (2/26)	MRC_Baseline	2022
Ministry of Statistics	Korean Standard Industry Classification	🎖 7th (7/311)	-	2022
Dacon	KLUE benchmark Natural Language Inference	🥇 1st (1/468)	🌐 KLUE NLI	2022
Dacon & AI Frenz	Python Code Clone Detection	🥉 3rd (3/337)	CloneDetection	2022
Dacon & CCEI Korea	Stock Price Forecast on KOSPI & KOSDAQ	🎖 6th (6/205)	elastic-stock-prediction	2021

**Dacon is Kaggle alike competition platform in Korea.

Generating dress outfit images based on given input text | 📄 Presentation

• Created training pipeline from VQGAN through DALLE
• Maintained versions of 1 million pairs image-caption dataset.
• Trained VQGAN and DALLE model from the scratch.
• Established live demo for the KoDALLE on Huggingface Space via FastAPI.

Implementation of Carlini et al(2020) Extracting Training Data from Large Language Models

• Accelerated inference speed with parallel Multi-GPU usage.
• Ruled out 'low-quality repeated generations' problem of the paper with repetition penalty and with ngram restriction.

Implementation of Shokri et al(2016) Membership Inference Attacks Against Machine Learning Models

• Prevented overfitting of shadow models' by adding early stop, regularizing with weight decay and allocating train/val/test datasets.
• Referenced Carlini et al(2021) to conduct further research on different types of models and metrics.
• Reproduced attack metrics as the following.

MIA ROC Curve CIFAR10	MIA ROC Curve CIFAR100

Paraphrasing tool with round trip translation utilizing T5 Machine Translation. | 🤗 KoQuillBot Demo & 🤗 Translator Demo

Implementation of Kasai et al(2020) Deep Encoder, Shallow Decoder: Reevaluating Non-autoregressive Machine Translation | 📄 Translation Output

• Composed custom dataset, trainer, inference code in pytorch and huggingface.
• Trained and hosted encoder-decoder transformers model using huggingface.

Extracting relations between subject and object entity in KLUE Benchmark dataset | ✍️ Blog Post

• Finetuned RoBERTa model according to RBERT structure in pytorch.
• Applied stratified k-fold cross validation for the custom trainer.

Sentence generation with given emotion conditions | 🤗 Huggingface Demo

• Finetuned KoGPT-Trinity with conditional emotion labels.
• Maintained huggingface hosted model and live demo.

Retrieved and extracted answers from wikipedia texts for given question | ✍️ Blog Post

• Attached bidirectional LSTM layers to the backbone transformers model to extract answers.
• Divided benchmark into start token prediction accuracy and end token prediction accuracy.

Corporate joint project for mathematics problems classification task | 📄 Presentation

• Preprocessed Korean mathematics problems dataset based on EDA.
• Maintained version of preprocessing module.

Created Emotional Instagram Posts(글스타그램) dataset | 📄 Presentation

• Managed version control for the project Github Repository.
• Converted Korean texts on image file into text file using Google Cloud Vision API.

Light-weight Neural Network for Optical Braille Recognition in the wild & on the book. | 🤗 Huggingface Demo

• Classified multi label one-hot encoded labels for raised braille patterns.
• Pseudo-labeled Natural Scene Braille dataset.
• Trained single stage object detection YOLO models for braille cell recognition.

Elimination based Lightweight Neural Net with Pretrained Weights | 📄 Presentation

• Constructed lightweight CNN model with less than 1M #params by removing top layers from pretrained CNN models.
• Assessed on Trash Annotations in Context(TACO) Dataset sampled for 6 classes with 20,851 images.
• Compared metrics accross VGG11, MobileNetV3 and EfficientNetB0.

Identifying 18 classes from given images: Age Range(3 classes), Biological Sex(2 classes), Face Mask(3 classes) | ✍️ Blog Post

• Optimized combination of backbone models, losses and optimizers.
• Created additional dataset with labels(age, sex, mask) to resolve class imbalance.
• Cropped facial characteristics with MTCNN and RetinaFace to reduce noise in the image.

Real-time desk posture classification through webcam | 📷 Demo Video

• Created real-time detection window using opencv-python.
• Converted image dataset into Yaw/Pitch/Roll numerical dataset using RetinaFace model.
• Trained and optimized random forest classification model with precision rate of 93%.

Overview for student life in foreign universities | ✈️ Website Demo

• 3400 Visitors within a year (2021.07 ~ 2022.07)
• 22000 Pageviews within a year (2021.07 ~ 2022.07)
• 3 minutes+ of Average Retention Time

• Collected and preprocessed 11200 text review data from the Yonsei website using pandas.
• Visualized department distribution and weather information using matplotlib.
• Sentiment analysis on satisfaction level for foreign universities with pretrained BERT model.
• Clustered universities with provided curriculum with K-means clustering.
• Hosted reports on universities using Gatsby.js, GraphQL, and Netlify.

Search-based exercise retrieval web service | 📷 Demo Video

• Built retrieval algorithm based on search keyword using TF-IDF.
• Deployed website using Docker, AWS RDS, AWS S3, AWS EBS
• Constructed backend using Django, Django ORM & PostgreSQL.
• Composed client-side using Sass, Tailwind, HTML5.

Federal Rate Prediction for the next FOMC Meeting

• Wrangled quantitative dataset with Finance Data Reader.
• Yielded metrics and compared candidate regression models for the adaquate fit.
• Hyperparameter optimization for the candidate models.

Get financial data of public companies involved in spinoff events on Google Spreadsheet | 🧩 Dataset Demo

• Wrangled finance dataset which are displayed on Google Sheets

Fixed torch version comparison fallback error for source repo of NVIDIA Research | ✍️ Pull Request

• Skills: torch, torchvision

Updated PostgreSQL initialization for "Quickstart: dockerizing django" documentation | ✍️ Pull Request

• Skills: Docker, docker-compose, Django

Predict the spread of COVID-19 in early stage after its entrance to country.

• Fixed existing errors on Github Repository.
• Wrote footnotes in both English and Korean.
• ±5% accuracy for one-day prediction.
• ±10% accuracy for 30-day prediction.

Don't miss concerts for your favorite artists with KakaoTalk Chatbot | 📷 Demo Video

• Created API server for KakaoTalk chatbot with Flask, Pymongo and MongoDB.
• Deployed the API server on AWS EC2.
• Visualized concert schedules on user's Google Calendar.
• Created / Updated events in Google Calendar.

Data Analysis and Machine Learning

• Data Analysis Library: pandas, numpy
• Deep Learning: pytorch, transformers
• Machine Learning: scikit-learn, gensim, xgboost

Backend

• Python / Django - Django ORM, CRUD, OAuth
• Python / FastAPI(uvicorn) - CRUD API
• Python / Flask - CRUD API

Client

• HTML / Pug.js
• CSS / Sass, Tailwind, Bulma
• JavaScript / ES6

Deployment

• Docker, docker-compose
• AWS EC2, Google Cloud App Engine
• AWS S3, RDS (PostgreSQL)
• AWS Elastic Beanstalk, CodePipeline;