Jingyang G.

Key Skills

Software

Other

Top Subject Matter

LLM application for enterprise knowledge Q&A (RAG, retrieval optimization, context management)

Enterprise Services – AI Agent Development & Workflow Automation

IT & Operations – Log Analysis & System Monitoring

Top Data Types

Computer Code Programming

Document

Video

Top Task Types

Question Answering

Classification

RLHF

Computer Programming/Coding

Function Calling

Fine-tuning

Text Generation

Object Detection

Entity (NER) Classification

Freelancer Overview

LLM Application R&D Intern at Baizhi Education (Tianjin). Experienced in building end-to-end AI systems, including RAG-based Q&A assistants, fine-tuned LLMs for vertical domains, and agent workflows. Core skills: RAG optimization (RRF+Bm25), LoRA fine-tuning, LangGraph agent orchestration, function calling, and SSE streaming. Projects include a sales agent platform (WeCom integration, multi-graph workflow) and a log analysis system (vLLM deployment, 50% troubleshooting time reduction). Bachelor of Software Engineering with coursework in ML and Python.

Labeling Experience

LLM Application R&D Intern — Baizhi Education Technology (Tianjin) Co., Ltd.

DocumentQuestion Answering

Built a RAG-based enterprise vertical domain knowledge assistant for Q&A using proprietary business data. Implemented hybrid retrieval combining vector search with BM25 reranking and secure recall rules to improve relevance and robustness. Developed multi-format document parsing and intelligent text chunking with context-overflow controls to support downstream model retrieval and answer generation. • Built retrieval pipeline with RRF fusion of vector and BM25 • Engineered SSE real-time streaming output for step-by-step responses • Implemented dual context limits (conversation rounds + token count) with cleanup strategy • Achieved ~85% recall and ~78% precision, reducing context overflow by ~80%

2025 - 2025

Education

Work History