Get early access
Back to episodes
#9over 1 year ago

Cracking the Prompting Interview

Ready to level up your prompting skills? Join us for a deep dive into advanced prompting techniques that separate good prompt engineers from great ones. We'll cover systematic prompt design, testing tools / inner loops, and tackle real-world prompting challenges. Perfect prep for becoming a more effective AI engineer.

Project details

Open in GitHub

Cracking the Prompting Interview

Ready to level up your prompting skills? Join us for a deep dive into advanced prompting techniques that separate good prompt engineers from great ones. We'll cover systematic prompt design, evaluation frameworks, and tackle real interview-style prompting challenges.

Video (1h23m - Available June 13, 2025 8 AM PST)

🎯 Key Takeaways

  • Use Indexes for URLs & Citations: Provide content with simple IDs (e.g., [SOURCE_1]) and have the LLM output these IDs. Map them back programmatically to improve accuracy and reduce token load.
  • Index-Based Diarization: For tasks like speaker diarization, have the LLM output the index of the dialogue turn and the identified speaker (e.g., {"dialogue_idx": 0, "speaker": "Nurse"}).
  • Context & "Escape Hatches" for Classification: Provide relevant context upfront and include an "Other" or "Unknown" category to handle ambiguity.
  • Reasoning via "Busted" JSON/Comments: Include LLM reasoning as comments or non-standard fields in structured output for easier debugging.
  • Natural Code Generation (in JSON): Generate code within Markdown-style backticks as a string field in JSON for higher quality output.
  • RTFP (Read The...Prompt!): Carefully review prompts for potential ambiguities that might confuse the LLM.

📝 Whiteboards

image

image

image

image

🚀 Running the Code

uv sync
uv run hello.py
uvx run baml-cli test

đź“– Resources

All episodes