Machine learning

Research on the open web, in feeds and in notebooks: pick the method that fits the question — pulse for what people say right now, deep-research for a…

Audit the vault's PARA and Zettelkasten-inspired Atomic Note organization, linked knowledge, journals, metadata, LLM Wiki quality, and bounded statistics.

Secret safety for AWS Secrets Manager, secret management, credentials, API keys, tokens, and passwords.

Access 50+ LLM models through AceDataCloud's unified chat APIs.

Harden code where untrusted input enters the system.

免費公開揭露資料偵察員(付費資料庫的免費對應+事件研究的事件源)。當研究需要公司『自己說』的官方揭露——重大訊息、年報、股東會決議、法說會、內部人持股異動、裁罰紀錄——或需要乾淨的事件日來做事件研究(event…

可重現性與複製包架構師(2026 頂刊資料編輯標準):把研究做成第三方能重跑、抵得住資料編輯審查的複製包(replication package)。涵蓋:可重現專案結構(raw/唯讀、碼與輸出分離、一鍵 master script)、計算環境捕捉(renv/conda、種子、版本)、Quarto…

研究架構圖（概念模型圖）產生器。把研究假說畫成投稿與口試等級的架構圖：中介模型 X→M→Y、調節模型、被調節的中介、序列中介，以及台灣商管論文最常用的『中介＋雙控制變數框』版式（每個依變數各一組控制變數，以虛線彎箭頭接入）。輸入一份 JSON 規格，輸出可再編輯的向量圖：SVG（零依賴、可用…

空間與地理資料分析架構師(區域研究 / hedonic 房價 / 不動產實證)。把帶有地址或經緯度的資料變成可分析、可畫圖、可發表的空間研究資料集。核心能力:(1)LLM 輔助地理編碼＋驗證退路——地址轉座標後,用行政區界/中心點/白名單比對抓出離群與幻覺,絕不讓 AI 亂猜的座標進資料;(2)H3…

LLM-as-a-judge rubric for code comments (forbidden, false, stale, narration, noise, keep).

ML model evaluation and experiment tracking -- LLM benchmarking with lm-evaluation-harness (MMLU, GSM8K, HumanEval, 60+ tasks), experiment tracking and…

LLM fine-tuning and alignment -- LoRA/QLoRA with PEFT, RLHF/DPO/GRPO with TRL, YAML-driven training with Axolotl.

Data science and analytics expertise for statistical analysis, machine learning pipelines, data governance, business intelligence, predictive modeling, and…

Analyze the most expensive users in AI observability and explain why they cost so much.

LLM and cloud cost awareness — model tiering, token budgets, right-sizing, and when a cheaper model suffices.

Improve AI application with evaluation-driven development.

Writing and changing code: Ponytail simplicity, grounded style rules, local and feature-wide systematic debugging, completion verification, architecture…

Configure and use Monica AI providers, chat, knowledge bases, RAG, runtime skills, and MCP in an application.

Authorized offensive and defensive security assessments, one router over per-surface playbooks: API, cloud/K8s, code audit & SAST review, database, digital…

Create high-quality, model-aware system prompts for any LLM (Claude, GPT, Gemini, open-source, etc.).

Layered timeout defense for cron-orchestrated agents calling external APIs (LLM CLI, HTTP, subprocess) — nested tiers, anti-patterns, headroom.

Answer a question from the personal knowledge base (LLM wiki) at ~/my-knowledge-base.

Select and run Python SDK verification with nox, Makefile targets, Ruff, mypy, pytest markers, sanity tests, type inference checks, and build checks.

Use for "interrogate", "adversarial review", "multi-model review", "challenge this", "stress test this code", "find blind spots", or "tear this apart".