Schultz D.

Key Skills

Software

Label Studio

Other

Top Subject Matter

Telecommunications (5G/6G)

network technical documents

professional multi-document scenarios

Top Data Types

Document

Text

Top Task Types

Prompt + Response Writing (SFT)

Diagnosis

Text Summarization

Freelancer Overview

Large Model Professional Domain Data Construction (AI training data & evaluation/rating). Brings 4+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Label Studio, Dify, and Other. Education includes Doctor of Philosophy, Xidian University (2029) and Bachelor of Science, Heilongjiang University (2024). AI-training focus includes data types such as Document and labeling workflows including Prompt + Response Writing (SFT), Diagnosis, and Text Summarization.

Labeling Experience

Large Model Professional Domain Data Construction (AI training data & evaluation/rating)

Label StudioDocumentPrompt + Response Writing (SFT)

Created high-quality instruction datasets requiring multi-document context, user constraints, and attached artifacts (e.g., PDFs, configs, logs) to support supervised fine-tuning. Designed evaluation rubrics to score model outputs for factual accuracy, reasoning coherence, safety, and domain alignment, then used structured feedback to improve dataset standards. Performed corpus cleaning and deduplication to raise data quality for training and evaluation workflows. • Multi-document instruction dataset construction • Rubric design and structured feedback for labeling quality • Data cleaning and deduplication for SFT/classification pipelines • Validation of LLM outputs across factuality, reasoning, safety, and domain adherence

2024 - Present

iGotHub - GitHub Trending Project Discovery Assistant (RAG-based content summarization)

OtherDocumentText Summarization

Built an AI assistant that recommends projects and performs tech-stack analysis with retrieval-backed context handling. Implemented RAG pipelines over GitHub content (READMEs, release notes, issue threads) to support grounded responses and technical summarization. Used retrieval context to structure outputs for downstream evaluation and usability. • Retrieval-augmented summarization over GitHub documents • Grounded responses using RAG over code/repo artifacts • Technical stack analysis and auto-summary generation • Assistance generation from README/release/issue threads

2023 - Present

LLM-Based Network Fault Diagnosis System (fault-case collection, annotation, normalization)

DocumentDiagnosis

Collected, annotated, and normalized network fault cases from syslogs, SNMP traps, and operator manuals to create labeled training data for an LLM-based diagnosis workflow. Developed and orchestrated a Dify-based system that used fine-tuned LLMs to produce root-cause analysis and recovery-plan generation in network-fault scenarios. Focused on ensuring labeled data quality through normalization of heterogeneous telecom records. • Annotation and normalization of fault cases (syslog/SNMP/manuals) • Training-data preparation for network fault diagnosis • Fine-tuned LLM orchestration for RCA and recovery plans • Dataset quality improvements via consistent labeling/normalization

2023 - 2024

Education

Work History