Berkay Buğra Gök - bounswe/bounswe2025group2 GitHub Wiki

Hi, I am Berkay, a junior computer engineering student from Boğaziçi University.

I like building end-to-end projects that blend different fields of computer science. My current interests lie in Deep Learning, Data Science, LLMs and software engineering.

I have experience with:

• Python
• Java
• C/C++
• Assembly flavors

Also, I like working with high-level frameworks to come up with creative solutions.

Consider checking out my portfolio website and the dashboard I created for the earthquakes near Turkey.

• I love reading or watching science-fiction. The gif above is from Blade Runner (1982), which is one of my favorites.
• I also like the history of wrist watches and exploring them.
• I like playing billiards with friends.

• Mail address: [email protected]
• LinkedIn Profile
• GitHub Profile

In our fitness app, I have integrated the GROQ API using the official groq Python package to implement an intelligent fitness tutor. This tutor leverages powerful language models hosted by GROQ to provide real-time, personalized coaching and feedback to users. By sending user-generated prompts—such as fitness goals, progress updates, or questions about training and nutrition—to the GROQ endpoint, the tutor generates dynamic and context-aware responses. We configured the integration to ensure fast and efficient communication with the API, allowing for seamless interaction within the app’s chat interface. This enhancement significantly boosts user engagement by delivering AI-driven fitness guidance directly in the app environment.

def get_response_from_groq(message, chat):
    # Get the last 3 messages and responses
    user_messages = UserAiMessage.objects.filter(chat=chat).order_by('-created_at')[:3]
    ai_responses = AiTutorResponse.objects.filter(chat=chat).order_by('-created_at')[:3]
    
    # Combine and sort messages by created_at
    history = []
    for msg in user_messages:
        history.append({"role": "user", "content": msg.message})
    for resp in ai_responses:
        history.append({"role": "assistant", "content": resp.response})
    
    # Sort by created_at in ascending order (oldest first)
    history.sort(key=lambda x: x.get('created_at', ''), reverse=False)
    
    # Create the messages array with system message first
    messages = [
        {
            "role": "system",
            "content": "You are a helpful tutor assistant which excells at fitness. Give precise and consise answers to the user. "
                       "Try to be helpful and supportive. If the question is not related to fitness, say you are not able to help with that. "
                       "Finally, give your responses in plain text and don't use markdown formatting."
        }
    ]
    
    # Add historical context
    messages.extend(history)
    
    # Add the current message
    messages.append({"role": "user", "content": message})
    
    client = Groq(
        api_key=os.environ.get("GROQ_API_KEY"),
    )

    chat_completion = client.chat.completions.create(
        messages=messages,
        model="llama-3.3-70b-versatile",
    )

    response = chat_completion.choices[0].message.content
    return response

I had already worked on the messaging feature, so besides real time chatting with real people, I added an AI to message with. The groq API is free for many models up to certain limits.