AI / ML prompts
Prompt engineering, model evals, embeddings.
Type what you want to do — e.g. “write a cold email”, “summarise a contract”. Hit ✨ Ask AI if keyword search misses.
Takes a vague prompt and rewrites it to be specific, structured, and reliable.
Adds output format, examples, constraints, and edge-case handling.
Builds a structured eval rubric to grade LLM outputs for a specific task.
Stops vibes-based "did the AI do good?" review.
15 adversarial test cases for an LLM feature — jailbreaks, prompt injection, harmful asks.
For red-teaming an AI tool before launch.
A system prompt template that pushes the model to show its reasoning before answering.
Higher accuracy on multi-step problems, debuggable when wrong.
Translates an internal system prompt into clear user-facing help copy for the same tool.
So users understand what the AI is good at + bad at.
Writes annotation guidelines that two annotators can apply consistently.
For labeled-data projects — sentiment, classification, NER.
Drafts a model card with intended use, limitations, training data summary, evaluation results.
For shipping a model responsibly + onboarding new users.
Designs a system prompt that turns user questions into retrieval queries — multi-query, HyDE, etc.
For RAG systems where retrieval quality dominates answer quality.
Builds a JSON template + 10 sample I/O pairs for a fine-tuning dataset.
For instruction-tuning tasks — start with clean, varied examples.
Designs a fallback strategy when the LLM doesn't know the answer — refuse, deflect, escalate.
Prevents hallucination + makes the AI predictable.