Đăng N.

Key Skills

Software

Don't disclose

Top Subject Matter

Document Layout Analysis (Vision-Language/Multimodal Document Understanding)

Retrieval-Augmented Generation for GraphRAG

Algorithms Education (Program Assignments and Assessment)

Top Data Types

Document

Image

Top Task Types

Fine-tuning

Data Collection

Freelancer Overview

Graduate Research Assistant. Brings 4+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Don't disclose. Education includes Bachelor of Science, Hanoi University of Science and Technology (HUST) (2026) and Secondary School Specialization Program (Mathematics), Ha Tinh High School for the Gifted (2022). AI-training focus includes data types such as Document, Medical, and DICOM and labeling workflows including Fine-tuning, Data Collection, and Evaluation.

Labeling Experience

Exchange Internship, West Japan Railway Company (Mar 2025 – Jun 2025)

Don't discloseData Collection

Researched and implemented hybrid retrieval pipelines for GraphRAG by combining BM25 keyword search with semantic retrieval. Built retrieval-centric components that support downstream model training and evaluation using mixed lexical and embedding-based signals. Although no explicit labeling is stated, the work supports preparing and curating training/evaluation data through retrieval. • Implemented hybrid retrieval using BM25 + semantic search. • Developed GraphRAG-style pipeline logic for knowledge retrieval. • Enabled selection of relevant text/graph context for model use. • Supported training or evaluation workflows via retrieval outputs.

2025 - 2025

AI Engineer Intern, FPT Smart Cloud (Apr 2024 – Aug 2024)

Don't discloseDocumentFine-tuning

Conducted research and large-scale experiments on document layout analysis using advanced vision architectures and document understanding pipelines. Fine-tuned state-of-the-art models for document understanding, while exploring modern LLM techniques such as RAG and function calling to improve performance. Focused on training workflows for vision and multimodal systems rather than manual labeling. • Worked on document layout analysis training pipelines. • Fine-tuned models including Vision Grid Transformer, DiT, YOLOv8, and YOLOv10. • Improved model performance by ~5%–7% mAP during experiments. • Explored RAG and function calling approaches as part of training/inference workflows.

2024 - 2024

Teaching Assistant, School of Information and Communication Technology - HUST (Sep 2023 – Feb 2024)

Don't disclose

Assisted as a teaching assistant by designing and preparing programming exercises and assessments for coursework evaluation. Covered core algorithm topics including recursion, dynamic programming, and pointers to support student practice and grading. While this is not classic data labeling, the role involved training/assessment activities related to AI/ML education practices. • Designed programming exercises and assessment content. • Evaluated student coursework performance on algorithm topics. • Provided guidance and assistance on course materials to students. • Supported learning through structured practice and feedback.

2023 - 2024

Education

Work History