This page serves as a comprehensive documentation hub for all projects and assignments completed during the Maven AI Engineering Bootcamp & Certificate program. Each week's work, including assignments, code repositories, deployment links, and learning reflections, will be organized here to track progress and showcase the skills developed throughout the bootcamp.
Render URL and GitHub repo links
Screenshots showing all 3 endpoints working
Health Check

Summarize

Sentiment

For each LLM endpoint, show your 3 prompt variations with example outputs and explain which worked best
Summarize Endpoint prompt variations
zero_shot prompt input and response:
Input Payload :
{
"text": "This is new to me",
"max_length": 1,
"prompt_type": "zero_shot"
}
Response:
{
"summary": "Unf"
}
few_shot prompt input and response:
Input Payload :
{
"text": "Hello AI !!!",
"max_length": 1,
"prompt_type": "few_shot"
}
Response:
{
"summary": "Greeting"
}
chain_of_thought prompt input and response
Input Payload :
{
"text": "Hello AI !!!",
"max_length": 1,
"prompt_type": "chain_of_thought"
}
Response:
{
"summary": "Main points"
}
meta prompt input and response
Input Payload :
{
"text": "Hello AI !!!",
"max_length": 1,
"prompt_type": "meta"
}
Response:
{
"summary": "Greetings"
}
Analyze Sentiment endpoint prompt variations
zero_shot prompt input and response:
Input Payload :
{
"text": "Hello AI !!!",
"max_length": 1,
"prompt_type": "zero_shot"
}
Response:
{
"sentiment": "positive",
"confidence_score": 0.9,
"explanation": "The use of 'Hello' along with multiple exclamation marks suggests enthusiasm and a friendly tone, indicating a positive sentiment."
}
few_shot prompt input and response
Input Payload :
{
"text": "Hello AI !!!",
"max_length": 1,
"prompt_type": "few_shot"
}
Response:
{
"sentiment": "neutral",
"confidence_score": 0.8,
"explanation": "The text is a greeting with no emotional content or sentiment expressed."
}
chain_of_thought prompt input and response
Input Payload :
{
"text": "Hello AI !!!",
"max_length": 1,
"prompt_type": "chain_of_thought"
}
Response:
{
"sentiment": "positive",
"confidence_score": 0.9,
"explanation": "The text conveys a positive sentiment through a friendly greeting and enthusiastic punctuation."
}
meta prompt input and response
Input Payload :
{
"text": "Hello AI !!!",
"max_length": 1,
"prompt_type": "meta"
Response:
{
"sentiment": "positive",
"confidence_score": 0.9,
"explanation": "The use of 'Hello' combined with multiple exclamation marks suggests a friendly and enthusiastic tone, indicating a positive sentiment."
}
Summary about prompt engineering and few of the techniques
Prompt engineering is all about providing more context or instructions on what specifically you want to achieve with the system.
Types of Prompting (focusing on the ones I looked into while working on the FastAPI week 1 project):
are to name a few.
zero-shot prompting but it depends on the complexity and knowledge of the task at hand and the tasks the model was trained to perform good on. In the summarize example above using zero-shot prompting, there was no detailed context provided in the input "This is new to me” resulting the model to respond unf - Unfamiliar Finetuning Examples which are used to investigate and control how LLMs hallucinate or generate plausible-sounding but factually incorrect responses. The same when tried with chain_of_thought prompt resulted in a response of Main Points