张龙

Key Skills

Software

Don't disclose

Top Subject Matter

Computer Vision / Traffic Violation Detection

Information Retrieval / LLM-based QA (RAG)

Top Data Types

Document

Top Task Types

Fine-tuning

Question Answering

Freelancer Overview

Intelligent Traffic Detection System based on YOLOv8 (Graduation Project). Brings 2+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Don't disclose and LangChain + DeepSeek-V4 Pro + FastAPI + Gradio. Education includes Bachelor of Science, Anhui Normal University (2026). AI-training focus includes data types such as Computer Code and Programming and labeling workflows including Fine-tuning and Question Answering.

Labeling Experience

RAG Knowledge Base QA System

Question Answering

You built a RAG knowledge-base QA system that indexes documents for semantic retrieval and question answering. The pipeline loads multi-format documents (PDF, Word, Markdown), performs semantic chunking, and creates vector indexes for retrieval. You implemented an SSE streaming QA API with multi-turn context and citation tracing to support explainable answers. • Document ingestion: PDF/Word/Markdown loading • Retrieval preparation: semantic chunking and vector indexing • Serving layer: FastAPI streaming QA with citations • Frontend: Gradio UI for upload/management/chat and one-click deployment

2026 - Present

Intelligent Traffic Detection System based on YOLOv8 (Graduation Project)

Don't discloseFine-tuning

You trained a YOLOv8-based vehicle detection model on a proprietary vehicle dataset to support real-time traffic violation recognition. You also independently trained a traffic light status detection model to improve the accuracy of intersection violation judgment. The work involved preparing and using labeled data to fine-tune and evaluate the models for detection performance. • Data used: proprietary vehicle dataset and traffic-light-related dataset • Labeling/annotation used for model training pipeline • Objective: detect vehicles and traffic states for violation decisions • Integration: connected model outputs into a downstream violation recognition system

2025 - 2026

Education

Work History