Prompt Engineering for Developers @ AWS Community Day Timișoara

My presentation from the AWS Community Day in Timişoara, Romania (April 10th, 2025).

## Prompt Engineering for Developers

Everyone talks about prompt engineering. Prompts seem like a magical way of communicating with LLMs. If you do not type a specific set of spells, LLM will hallucinate a rogue response.

But what are prompts, and how do they work? This talk will demystify the magic behind prompts and teach you how to use them in practice when building applications (both as a developer tool and as a part of the application you are building).

Transcript

1. It's hard to give clear instructions
2. SMS Goeben August 1914, the Mediterranean
3. TL;DR, Wait until war is declared, and destroy the Goeben Winston Churchi!, head of Admiralty, July 31, 1914
4. SMS Goeben
5. SMS Goeben
6. TL;DR, Wait until war is declared, and destroy the Goeben "… do not at this stage be brought to action against superior forces." Winston Churchi!, head of Admiralty, July 31, 1914
7. SMS Goeben
8. SMS Goeben
9. For Churchi!: Superior forces = Austrian battleships
10. I am sure that your unclear instructions never caused major disruption of someone's fleet*, but you definitely have many stories * the Goeben did that to the Russian fl"t in the Black Sea during WW1
11. Let's talk about giving clear instructions!
12. Prompt Engineering for developers
13. Prompt Engineering , and other people who love talking to computers for developers
14. Human communication is complicated
15. Luckily, it's a bit more clear how Large Language Models (LLMs) work
21. Source: https://platform.openai.com/tokenizer
22. Prompts: - Prompts are instructions - You te! an LLM what you want, and it tries to reply based its training and your instructions - more clear instructions = better reply - LLM always answer, but not always based on truth
23. Slobodan Stojanovic cofounder/CTO @ Vacation Tracker AWS Serverless Hero @slobodan_
24. Prompts are just instructions
25. So, how LLMs work?
26. You give your instructions
28. You get some unexpected wisdom or hallucination
30. Tokens and vectors
31. @slobodan_
32. Anatomy of a prompt
33. The prompt is a set of textual instructions that fit LLM's context and other limitations
34. "Who is faster: Godzi!a or T-Rex?" A valid prompt
35. "Write a 500-word article about the bad influence of Amazon's RTO policy on Lambda cold starts" Also valid prompt
38. But LLMs are products that, like most other products, evolve with user needs and requests
39. System prompts
41. Some parts of your instructions might be more important than other or you might want to make them repeatable
42. Usage via the API gives you additional superpowers
43. Repeatability
44. (simulated) conversations
45. more control
47. Successful prompts require writing
48. I know you hate writing
49. But you wrote thousands of lines of JavaScript or Python last month
50. That's also just a bunch of text and instructions
51. 7 Habits of highly effective prompters
52. Five simple tricks for better prompting results
53. 1. Hint the beginning of the answer
54. We often need a response from an LLM to follow the specific structure we defined
55. User: // Some long instructions But always reply with valid JSON and nothing else! Here's your JSON: ```json { "some": "JSON", Assistant:
56. I SAID JSON ONLY!!! Works, sometimes
57. But there's something else you can do!
58. Write the beginning of the reply in the API request!
59. User: // Your instructions Answer with valid JSON and nothing else. { " Assistant: System: // Your system prompt
60. User: // Your instructions Answer with valid JSON and nothing else. { " Assistant: System: // Your system prompt some": "valid", "JSON": true }
61. 2. Give examples
62. We can use the same idea for one more trick
63. ChatGPT and Claude often don't do the task perfectly immediately. It takes some back and forth before you get the desired outcome.
64. But we often try to do just one direct command via the API
65. There's a fancy name for a single direct co#and: Zero-Shot Prompting
66. Because, there's another popular fancy name: Few-Shot Prompting
67. "Great product, 10/10": {"label": "positive"} "Didn't work very well": {"label": "negative"} "Super helpful, worth it":
68. "Great product, 10/10": {"label": "positive"} "Didn't work very well": {"label": "negative"} "Super helpful, worth it": {"label": "positive"}
69. Like a mini "in-prompt" training
70. But you can also use this differently (and for more complicated use cases)
71. User: Assistant: "Great product, 10/10" {"label": "positive"} User: Assistant: "Didn't work very we!" {"label": "negative"} User: Assistant: "Super helpful, worth it" System: // Your system prompt
72. User: Assistant: "Great product, 10/10" {"label": "positive"} User: Assistant: "Didn't work very we!" {"label": "negative"} User: Assistant: "Super helpful, worth it" {"label": "POSITIVE"} System: // Your system prompt
74. 3. Think step by step
75. The next trick would work good with humans, too
76. I want you to create this unreasonable feature! And I want it now!
77. Let's think step by step. How would that feature help our users?
78. But in reality, it would probably just give you a new Linkedin employment status
79. #opentowork
80. Luckily, LLMs have no emotions
81. Adding "Let's think step by step" or a similar phrase uses more tokens for the response, but it often gives you better results
82. From OpenAI O1 preview announcement: "We've developed a new series of AI models designed to spend more time thinking before they respond."
83. From OpenAI O1 preview announcement: "They can reason through complex tasks and solve harder problems than previous models in science, coding, and math."
84. 4. Use tools
85. LLMs sucks at math and some other tasks
86. Tricks above might improve it a bit but often not enough
87. But all major LLMs support tools
88. So, instead of this:
89. User: Assistant: "Do some complex calculations" Brand new ha!ucination System: // Your system prompt
90. You can do this:
91. User: Assistant: Do some complex calculations and give me an analysis. I need an answer from calc tool, here are the arguments to pass it System: // Your system prompt
92. User: Assistant: Do some complex calculations and give me an analysis. I n"d an answer from calc tool, here are the arguments to pass it User: Assistant: // Ca!s JS function and returns result Here's the analysis System: // Your system prompt
93. Tools allow an LLM to ask you for some help
94. Tools power AI Agents!
95. But, tools are just simple functions! (JavaScript, Python, PHP, or whatever you want)
96. 5. Ask an LLM to improve the prompt
97. Can an LLM write a prompt?
98. A prompt is just a textual command. It can review and improve it!
99. Guess what? It works even better if you provide clear instructions!
100. Commander's Intent
101. Commander's Intent: - Purpose: Why personnel must complete the assignment. - Task: What the objective or goal entails. - End state: How the result should look.
102. You can try something like this: Here's my prompt: ${INITIAL PROMPT } Rewrite it to fo!ow the Co#ander's Intent Statement. Ask for missing details.
103. "Considerations in Communicating intent" Sources of Power: How People Make Decisions Book by Gary Klein
104. "There are seven types of information that a person could present to help the people receiving the request to understand what to do" Sources of Power Book by Gary Klein
105. 1. The purpose of the task (the higher level goals) 2.The objective of the task (an image of desired outcome) 3.The sequence of steps in the plan 4.The rationale for the plan 5.The key decisions that may have to be made 6.Antigoals 7.Constrains and other considerations
106. "A! seven types of information are not always necessary." Sources of Power Book by Gary Klein
107. @slobodan_ Summary
108. @slobodan_ • Hint the beginning of an answer, LLM will follow • Give examples (few-shot prompting) • Ask an LLM to think step by step • Help an LLM with tools (tools are just functions) • Ask an LLM to help you improve the prompt
109. @slobodan_
110. @slobodan_ https:/ /slobodan.me/posts/5-prompt-engineering-tips-for-developers/