Wang Lingjie
Community@Lingjie-wang · Qingdao,Shandong
A CS Undergraduate in SDU
Agent Skills by Wang Lingjie
Showing 91 vetted skills indexed across 1 GitHub repositories.
Lingjie-wangautorl
Plans and executes evidence-backed AutoRL workflows using Codex or Claude Code as executor.
Lingjie-wangautorl-skill-workflow-v2
Orchestrates file-backed AutoRL workflows from environment intake through guarded training and evaluation.
Lingjie-wangpaper-poster-html
Generates print-ready academic conference posters as HTML/CSS with measurement-driven quality gates.
Lingjie-wangmeta-apply
Applies staged skill-corpus patches after fresh cross-model jury review and human approval.
Lingjie-wangproof-checker
Verifies and fixes LaTeX mathematical proofs via cross-model adversarial review with audit reports.
Lingjie-wangpaper-talk
Generates conference talk slides, speaker notes, and audit reports from a compiled academic paper.
Lingjie-wangpaper-write
Drafts LaTeX academic papers section by section from a research outline and narrative report.
Lingjie-wangpaper-poster
Redirects poster generation requests to the paper-poster-html pipeline.
Lingjie-wanginterview-cheatsheet
Generates long-form Chinese ML/LLM interview cheat sheets with formulas, PyTorch code, and tiered questions.
Lingjie-wangwriting-systems-papers
Provides paragraph-level structural blueprints for 10-12 page systems conference papers.
Lingjie-wangqzcli
Manage GPU compute jobs on the Qizhi platform using a kubectl-style CLI.
Lingjie-wangspecification-writing
Writes complete patent specification sections from claims and invention disclosure documents.
Lingjie-wangproof-writer
Writes rigorous mathematical proofs for ML/AI theory with feasibility triage and dependency maps.
Lingjie-wangpaper-claim-audit
Verifies every numeric claim in a research paper against raw result files using a zero-context cross-model reviewer.
Lingjie-wangresubmit-pipeline
Orchestrates text-only resubmission of a polished paper to a new venue under frozen bibliography and structure constraints.
Lingjie-wangserverless-modal
Runs GPU training, inference, and batch workloads on Modal serverless cloud.
Lingjie-wangformula-derivation
Structures scattered research notes into coherent formula derivation packages with explicit assumptions and status.
Lingjie-wangtraining-check
Monitors WandB training metrics periodically to detect NaN, divergence, and idle GPUs.
Lingjie-wangmeta-optimize
Analyze ARIS usage logs and propose optimizations to SKILL.md files and workflow defaults.
Lingjie-wangpatent-novelty-check
Assess patent novelty and non-obviousness of an invention against prior art references.
Lingjie-wangvast-gpu
Provisions, configures, and destroys vast.ai GPU instances for ML training workloads.
Lingjie-wangresult-to-claim
Evaluates experiment results against intended research claims and routes to next actions.
Lingjie-wangfigure-spec
Generate deterministic architecture, workflow, and pipeline diagrams from JSON specs into editable SVG.
Lingjie-wangmermaid-diagram
Generate Mermaid diagram code with CLI verification and rendered PNG output.
Frequently Asked Questions About Wang Lingjie
FAQPage SchemaWhat tasks can I accomplish with Wang Lingjie's skill registry?▼
You can run the full research lifecycle: discover and refine ideas, plan and execute GPU experiments, audit claims and citations, write and compile LaTeX papers, build slides and posters, draft patents and grant proposals, and orchestrate AutoRL environment integration and training.
Who is the target user for these skills?▼
ML researchers, CS graduate students, and academic authors preparing NeurIPS/ICML/ICLR/OSDI submissions, plus inventors filing CN/US/EP patents and engineers running multi-seed training jobs on Modal, Vast.ai, or the Qizhi platform.
How do the AutoRL skills work in practice?▼
The chain starts with rl-task-clarifier to produce a task card, then rl-evidence-retrieval, rl-env-integrator, and rl-env-verifier establish a tested environment, and rl-framework-implementer builds training code under approval gates, coordinated by the autorl orchestrator skills.
What external services and dependencies do these skills require?▼
Skills invoke Codex or Claude Code executors via MCP, plus arXiv, Semantic Scholar, OpenAlex, Exa, Gemini, and DeepXiv for retrieval; WandB for training monitoring; Overleaf sync; Feishu notifications; and GPU backends including Modal, Vast.ai, and qzcli.
Are these skills free to use?▼
The skills are published openly in the public GitHub registry under Lingjie-wang at no cost, though several depend on third-party paid services such as OpenAI-compatible review endpoints, Exa search, and paid GPU rental on Vast.ai or Modal.