La collecte non autorisée de données, y compris les données personnelles des utilisateurs fournies par ce service, sans le consentement de chaque sujet de données, est interdite et refusée. Veuillez noter que la collecte de données publiques sans autorisation au moyen de dispositifs techniques tels que le crawling peut entraîner des sanctions pénales en vertu de la loi sur la protection des informations personnelles.
CS & Math student @ KAIST | Rise Global Winner’22 | AFS Global STEM Scholar’20
Carrière
Publications
Résumé de carrière par IA
Les informations sur la carrière de ISMAYILZADA JAVAD sont encore insuffisantes pour générer un résumé. Le résumé de carrière s'affichera une fois que plus d'informations auront été ajoutées.
Carrière
AI engineer
SmoreTalk Co., Ltd. · 정규직
2025년 6월 - 2025년 8월 · 3개월
Worked at SmoreTalk Co., Ltd. one of 7 companies selected by Google AI Academy. As an AI engineer, worked on development of generative AI workflows and AI agents for SmoreTalk's product, including prompt-based generation, inpainting, and model integration. Developed over 15 complex workflows including restoration, upscaling, 3D asset, consistent character, style and product related workflows using ComfyUI, Cog, FastAPI tools. Deployed and scaled workflow services for 15,000+ users, optimizing size and speed in cloud GPUs (NHN, Lambda Labs, Replicate).
As Undergraduate Research Intern at KAIST, co-authored a paper accepted to NAACL 2025 (C3NLP) and received Outstanding Paper Award, titled “WHEN TOM EATS KIMCHI: Evaluating Cultural Awareness of MLLMs in Mixed-Cultural Contexts.” Designed and implemented a fully automated pipeline to evaluate cultural biases in multimodal large language models (MLLMs). Built components for web scraping, image filtering, and duplication removal. Automated image masking using segmentation models (SAM) and synthesized ethnic variations with inpainting APIs. Created MIXCUBE, a benchmark of 2.5k culturally and ethnically diverse images across five countries, three categories enabling quantitative evaluation of MLLM cultural awareness.
Voir plus
학부생 연구원 UG Researcher
정규직
2024년 2월 - 2024년 7월 · 6개월
Working on the “ Personalized Large Language Models Guided by Weak Models for User Networks” project. Read different papers on LLM, and personalization techniques, learned/implemented Transformer, BERT, ELMo architectures, and experimented with different Prompt Engineering techniques for LLM.
Rise Residental Summit, London & Oxford, United Kingdom · 정규직
2023년 7월 - 2023년 8월 · 2개월
Was part of AI circle, joined on webscrapping, AI & Human Operating systems, application of AI & ChatGPT workshops. Presented capstone project involving investment networks with content based filtering algorithm.
Post-graduation, provided Olympiad lectures to junior team members, trianed them for selection rounds for Olympiads, offered lectures in online and in-person camps, and prepeared them for upcoming olympiads.
BP Global STEM Academies, Online Changemaker Curriculum · 정규직
2020년 3월 - 2021년 3월 · 1년 1개월
Got selected into this BP and American Field Service sponsored program to study University of Pennsylvania’s online modules on UN’s Sustainable Development Goals. Created a computerized system to help deliver groceries to the bedridden, built a beta model for donations, recruitment, and delivery paths. Received an Advanced Certificate on Global Citizenship for Social Impact for this project (7-8 hrs/week)
Did and exchnage semester at Technical University of Berlin, took “ Mathematical Introduction to Machine Learning”, “Machine Intelligence I: Supervised Methods”, “Human-Computer Interaction (HCI)”, “Continuous Software Engineering”, “Computer Graphics I”, “Ethics of AI"
Built a pipeline for synthesizing culturally diverse images using Stable Diffusion inpainting. Integrated image scraping, face detection, and race-aware generation to prepare datasets for VQA tasks. Structured for food, clothes, and festival categories.
프로젝트
microservices-demo
Technische Universität Berlin · 2024년 10월 - 2025년 3월 · 6개월
Extended Google’s Online Boutique with a multilingual microservice. Deployed it locally on Minikube with Skaffold & Ansible automation. Enabled GitHub Actions for CI/CD. Built as part of TU Berlin coursework on Continuous Software Engineering.
프로젝트
Classification of the images in CIFAR10 dataset into 10 categories using CNN
2024년 3월 - 2024년 4월 · 2개월
Built and compared CNN architectures (MLP, ConvNet, ResNet, Inception) using PyTorch on CIFAR-10. Trained with data augmentation and SGD. Logged results via TensorBoard, tracked accuracy vs. params, and saved best checkpoints per model.
프로젝트
Image Segmentation using Fully Convolutional Network (FCN)
2024년 3월 - 2024년 4월 · 2개월
Implemented semantic segmentation using FCN32/FCN8 in PyTorch with pretrained VGG16. Trained on PASCAL VOC dataset. Achieved ~88% pixel accuracy. Added optional DenseCRF post-processing to refine results and logged metrics via TensorBoard.
프로젝트
ml-from-scratch
2024년 2월 - 2024년 3월 · 2개월
Implemented ML algorithms from scratch using NumPy and PyTorch: Lasso/Ridge regression, K-Means, GMM (EM), SVMs (with QP), and RL in a custom Pygame gridworld. Focused on math, optimization, and visual intuition—no high-level ML libraries used.
