Skip to content

busdel/YZ-Bootcamp-Grup7

Repository files navigation

🇹🇷 Türkçe🇬🇧 English


🇹🇷 Türkçe Versiyonu

🫀 HeartHelper – Yapay Zekâ Destekli Kalp Sağlığı Asistanı

Google Yapay Zeka ve Teknoloji Akademisi – Mezuniyet Bootcamp / Grup 7
Takım İsmi: Grup-7


👥 Takım Üyeleri

  • Neslişah Ebral Durdu – Scrum Master
  • Sıla Carus – Product Owner
  • Büşranur Delice – Developer
  • Ozan İnaltekin – Developer
  • Yavuz Selim Özdemir – Developer

🌐 Proje Tanımı

HeartHelper, kalp ve damar hastalıklarıyla yaşayan bireylerin tedavi süreçlerini daha etkili, bilinçli ve kişiselleştirilmiş şekilde yönetmelerine yardımcı olmak amacıyla geliştirilen yapay zekâ destekli dijital bir sağlık asistanıdır.

Hem Türkçe hem de İngilizce olarak geliştirilen bu web tabanlı uygulama, ilaç kullanımı, egzersiz planı ve beslenme gibi konularda güvenilir rehberlere dayalı, hızlı ve kişiye özel bilgi sunar.

Retrieval-augmented generation (RAG) teknolojisi ile çalışan sohbet ekranı sayesinde kullanıcılar sorularını yazabilir ve saniyeler içinde bilimsel kaynaklara dayanan yanıtlar alabilir.

Metin tabanlı ve kullanıcı dostu arayüzü, düşük dijital okuryazarlığa sahip bireylerin bile kolayca kullanabilmesini sağlar.


🔗 Canlı Uygulama

👉 Uygulamayı Deneyin


✨ Temel Özellikler

  • 🤖 Yapay Zekâ Destekli Soru-Cevap: RAG teknolojisiyle güvenilir ve anlık yanıtlar.
  • 📚 Güncel ve Doğrulanmış İçerik: Sağlık Bakanlığı ve uzman kuruluşlara dayalı bilgi altyapısı.
  • 🧬 Kişiselleştirilmiş Öneriler: Kullanıcıya özel diyet, egzersiz ve ilaç bilgilendirmeleri.
  • 🖥️ Erişilebilir Arayüz: Dijital okuryazarlığı düşük bireyler için sade tasarım.
  • ⏱️ 7/24 Destek: Sağlık hatırlatıcıları ve acil durum yönlendirmeleri.
  • 🔄 Ölçeklenebilir Mimari: Diyabet, KOAH ve hipertansiyon gibi hastalıklara uyarlanabilir.
  • 🔐 Veri Güvenliği: KVKK ve GDPR uyumlu veri koruma.

🎯 Hedef Kitle

  • 18 yaş ve üzeri kalp-damar hastalığı olan bireyler
  • Kronik hastalığı olan genç, orta yaşlı ve yaşlı yetişkinler
  • Hasta yakınları ve bakıcılar
  • Sağlık profesyonelleriyle sınırlı iletişim kurabilen bireyler
  • Dijital sağlık çözümlerine ilgi duyan kullanıcılar
  • Sağlık kurumlarıyla iş birliği yapmak isteyen kuruluşlar (hastaneler, ASM’ler, STK’lar, kamu kurumları)

📌 Notlar

HeartHelper, yapay zekâ gücüyle toplumsal sağlık okuryazarlığını artırmayı hedefleyen sosyal bir inovasyon projesidir. Ölçeklenebilir yapısı ve iki dilli tasarımıyla, erişilebilir ve güvenilir dijital sağlık çözümleri için örnek bir modeldir.


Product Backlog URL

Trello Board - Sprint 1


English Explanation

🫀 HeartHelper – AI-Powered Cardiovascular Health Assistant

YZ Bootcamp – Google Artificial Intelligence and Technology Academy / Team 7
Team Name: Group-7


👥 Team Members

  • Neslişah Ebral Durdu – Scrum Master
  • Sıla Carus – Product Owner
  • Büşranur Delice – Developer
  • Ozan İnaltekin – Developer
  • Yavuz Selim Özdemir – Developer

🌐 Project Overview

HeartHelper is an AI-powered digital health assistant designed to help individuals manage cardiovascular diseases in a more effective, informed, and personalized manner.

Developed as a bilingual web-based application (available in both Turkish and English), HeartHelper provides quick, accurate, and accessible medical guidance on medication use, exercise plans, and nutrition — based on up-to-date clinical guidelines.

Using retrieval-augmented generation (RAG) technology, the assistant delivers instant answers to users’ questions via a chat interface, ensuring responses are backed by reliable medical sources.

With its text-based and user-friendly interface, HeartHelper is tailored for users of all digital literacy levels, from young adults to older individuals.


🔗 Live Demo

👉 Click to try the app


✨ Key Features

  • 🤖 AI-Powered Q&A System: Get instant, reliable answers using RAG technology.
  • 📚 Verified & Up-to-Date Content: Based on guidelines from authorities like the Ministry of Health and medical societies.
  • 🧬 Personalized Recommendations: Customized suggestions for exercise, diet, and medications.
  • 🖥️ Accessible Interface: Designed for all users, including those with limited digital literacy.
  • ⏱️ 24/7 Support: Continuous assistance and health reminders.
  • 🔄 Scalable Architecture: Easily adaptable to other chronic diseases (e.g., diabetes, COPD, hypertension).
  • 🔐 Data Security: Fully compliant with GDPR and KVKK standards.

🎯 Target Audience

  • Adults (18+) living with cardiovascular diseases
  • Individuals managing chronic conditions
  • Caregivers and family members
  • Users with limited access to healthcare professionals
  • Patients interested in digital health tools
  • Institutions (hospitals, NGOs, public agencies) seeking collaboration

📌 Notes

HeartHelper is a social innovation project aimed at improving public health literacy through the power of AI. With its scalable infrastructure and bilingual design, it serves as a model for inclusive, accessible, and medically sound digital healthcare solutions.


Sprint 1

  • Sprint içinde tamamlanması tahmin edilen puan: 35 Puan

  • Proje yönetim aracı olarak Trello kullanılmasına karar verildi.

  • Puan tamamlama mantığı: Toplamda proje boyunca tamamlanması gereken ortalama 100 puanlık backlog bulunmaktadır. 3 sprint'e bölündüğünde ilk sprint'in en azından 35 ile başlaması gerektiğine karar verildi. 1. sprintte Bilgi Tabanı Oluşturulacağı için 35 puan hedeflenmiştir, 2. sprintte Soru-Cevap Modülü hazırlanacağından 25-35 puan, 3. sprintte Uygulama Arayüzü ve Yayına Hazırlık süreçleri olacağı için 25-35 puanlama arası hedef konulmuştur.

  • Daily Scrum: Daily Scrum toplantılarının zamansal sebeplerden ötürü Whatsapp üzerinden yapılmasına karar verilmiştir. Daily Scrum toplantısı örneği odt olarak Readme'de tarafımızdan paylaşılmaktadır: 📄 Sprint 1 Daily Scrum Chats

  • Sprint board update: Sprint board screenshotları:

    image

    image

  • Ürün Durumu: Ekran görüntüleri:

    image

  • Sprint Review:

    • Veri seti seçildi
    • Chatbot kullanıcı akışı Trello üzerinde oluşturuldu
    • PDF/Web içerik yükleme ve metin parçalayıcı modül geliştirildi
    • Google API ile embedding üretildi
    • FAISS indeksleme ve kayıt/yükleme fonksiyonları tamamlandı
    • Projemiz için veri arama ve soru-cevap altyapısının en kritik kısmı olan bilgi tabanı başarıyla oluşturuldu.
    • Kodlar ve çıktılar tekrar kullanılabilir; yeni PDF eklenirse süreç kolayca tekrarlanabilir.
    • Sonraki adımda, bu bilgi tabanından akıllı arama veya chatbot uygulamaları geliştirilebilir.

Toplam 35 SP planlandı ve başarıyla tamamlandı.
Proje altyapısı Sprint 2’ye hazır.

  • Sprint Review katılımcıları: Neslişah Ebral Durdu,Sıla Carus,Büşranur Delice,Ozan İnaltekin,Yavuz Selim Özdemir

  • Sprint Retrospective:

    • Takım içindeki görev dağılımıyla ilgili düzenleme yapılması kararı alınmıştır.
    • Takım üyelerinin gelecek sprintlerde panoyu daha aktif kullanması gerektiğine vurgu yapılmıştır.
    • İkinci sprintte kullanıcıdan doğal dilde soru alma arayüzü yapılmasına karar verildi.
    • Soru için embedding üretme ve FAISS ile en yakın içerikleri bulmaya karar verildi.
    • Gemini LLM (örneğin gemini-pro) ile prompt oluşturma ve yanıt üretilmeye karar verildi.
    • Yanıtların kullanıcıya sunulmasına karar verildi. (streamlit veya terminal tabanlı)
    • Test ve doğruluk kontrolü yapılmaya karar verildi.
English Explanation
  • Estimated story points to be completed within the sprint: 35 Points

  • Trello was chosen as the project management tool.

  • Point allocation logic: The project backlog consists of approximately 100 story points to be completed overall. Divided into 3 sprints, it was decided that the first sprint should start with at least 35 points. Since the Knowledge Base will be created in Sprint 1, the target was set to 35 points. Sprint 2 will focus on the Question-Answer Module with a target of 25-35 points, and Sprint 3 will cover the Application Interface and Deployment preparation with a target range of 25-35 points.

  • Daily Scrum: Due to timing constraints, Daily Scrum meetings are held via WhatsApp. An example of Daily Scrum chats is shared in our readme as an ODT file:📄 Sprint 1 Daily Scrum Chats

  • Sprint Review:

    • Dataset was selected
    • Chatbot user flow was created on Trello
    • PDF/Web content upload and text chunking module were developed
    • Embeddings were generated using Google API
    • FAISS indexing and save/load functions were completed
    • The critical part of our project infrastructure, the knowledge base for data search and question-answering, was successfully built.
    • Codes and outputs are reusable; if new PDFs are added, the process can be easily repeated.
    • Next steps include developing intelligent search or chatbot applications based on this knowledge base.

A total of 35 story points were planned and successfully completed.
The project infrastructure is ready for Sprint 2.

  • Sprint Review participants: Neslişah Ebral Durdu, Sıla Carus, Büşranur Delice, Ozan İnaltekin, Yavuz Selim Özdemir

  • Sprint Retrospective:

    • It was decided to reorganize task distribution within the team.
    • Team members were encouraged to use the Trello board more actively in upcoming sprints.
    • It was decided to develop a natural language question input interface for users in Sprint 2.
    • Generating embeddings for questions and retrieving the closest contents using FAISS was planned.
    • It was decided to create prompts and generate answers using the Gemini LLM (e.g., gemini-pro).
    • Delivering the answers to users (via Streamlit or terminal-based interface) was decided.
    • Testing and accuracy verification were planned.

Sprint 2

  • Sprint Notları:

    • Doğal dilde soru alabilmek için kullanıcı arayüzü geliştirildi (Streamlit).
    • Soru cümlelerinden embedding üretmek için Google Embedding API kullanıldı.
    • En alakalı içerikleri bulmak için FAISS vektör arama algoritması kullanıldı.
    • Alakalı içeriklerden cevap üretmek için Gemini-Pro LLM entegre edildi.
    • Uygulama arayüz tasarımı için Figma kullanıldı.
    • Kullanıcıdan yaş, cinsiyet gibi bilgiler almak için giriş formu geliştirildi.
    • İngilizce dil desteği sağlamak için çok dilli arayüz desteği eklendi.
    • Uygulamanın tanıtımı için kısa bir video hazırlandı (demo).
    • Sprint planlama ve günlük scrum toplantıları düzenli olarak yapıldı.
    • Trello üzerinden görev paylaşımı yapıldı ve ekip takibi sağlandı.
  • Sprint içinde tamamlanması tahmin edilen puan: 33 Puan

  • Puan tamamlama mantığı: Toplamda proje boyunca tamamlanması gereken ortalama 100 puanlık backlog bulunmaktadır. 1. sprintte Bilgi Tabanı Oluşturulacağı için 35 puanlık hedef tamamlanmıştır, 2. sprintte Soru-Cevap Modülü hazırlanıp 33 puan tamamlanmıştır, 3. sprintte Uygulama Arayüzü ve Yayına Hazırlık süreçleri ve test süreçleri olacağı için 25-35 puanlama arası hedef konulmuştur.

  • Daily Scrum: Daily Scrum toplantılarının zamansal sebeplerden ötürü Whatsapp üzerinden yapılmasına karar verilmiştir. Daily Scrum toplantısı örneği odt olarak Readme'de tarafımızdan paylaşılmaktadır: 📄 Sprint 2 Toplantı Notları

  • Product Backlog URL: Trello Board - Sprint 2

  • Sprint board screenshotları:

    image image image
  • Ürün Durumu: Ekran görüntüleri:

    image image
  • Sprint Review:

    Sprint 2 kapsamında, projemizin temel yapı taşlarından biri olan soru-cevap modülünü işlevsel hale getirmek için çeşitli teknik ve arayüz geliştirmeleri gerçekleştirdik. Bu sprintin ana hedefi, kullanıcının doğal dilde sorduğu sorulara sistemin anlamlı ve bağlamsal olarak doğru yanıtlar üretebilmesini sağlamaktı. Aşağıda sprint boyunca yapılan çalışmalar detaylı şekilde özetlenmiştir:

    • Soru alma ve embedding işlemi: Kullanıcının arayüz üzerinden yönelttiği doğal dildeki sorular, belirlenen bir Google metin embedding modeli ile vektörel forma dönüştürüldü. Bu dönüşüm sayesinde soruların anlamına en yakın içeriklerin bulunabilmesi mümkün hale geldi.

    • Vektör indeksleme ve arama: FAISS kütüphanesi kullanılarak içerik verileri vektör uzayında indekslendi. Kullanıcının sorusuna en yakın içerikleri hızlı ve etkili şekilde bulmak amacıyla bu yapı kuruldu. FAISS verileri disk üzerine yazıldı ve daha sonra tekrar yüklenebilir olacak şekilde yapılandırıldı.

    • LLM entegrasyonu: FAISS'ten elde edilen en alakalı içerikler, bir prompt şablonuna yerleştirilerek Google Gemini Pro modeline gönderildi. Bu model, verilen içeriklere göre kullanıcı sorusuna özgün ve anlamlı cevaplar üretti.

    • İngilizce dil desteği: Uygulamanın farklı kullanıcı profillerine hitap edebilmesi amacıyla İngilizce dil seçeneği eklendi. Giriş ekranında dil tercihi yapılarak, ilgili dilde cevap alınması sağlandı.

    • Kullanıcı bilgileri: Sistem, kullanıcıdan yaş, cinsiyet ve benzeri temel bilgileri alacak şekilde genişletildi. Bu bilgiler, gelecekte daha kişiselleştirilmiş yanıtlar üretmek için kullanılabilecek altyapının bir parçasını oluşturuyor.

    • Arayüz tasarımı: Figma kullanılarak kullanıcı arayüzü için modern ve kullanıcı dostu tasarımlar geliştirildi. Arayüz, sade ancak işlevsel olacak şekilde planlandı.

    • Tanıtım çalışması: Uygulamanın temel işleyişini ve kullanıcı deneyimini yansıtan kısa bir tanıtım videosu hazırlandı.

    • Scrum süreci: Sprint boyunca günlük Scrum toplantıları düzenli olarak yapıldı. Scrum Master, görev takibini ve ekip koordinasyonunu sağladı. Görevler Trello üzerinden yönetildi ve iş bölümü etkin şekilde yürütüldü.

  • Sprint Review katılımcıları: Neslişah Ebral Durdu,Sıla Carus,Büşranur Delice,Ozan İnaltekin,Yavuz Selim Özdemir

  • Sprint Retrospective:

    • Uygulamanın deploy sürecinin tamamlanmasına karar verildi.
    • Logo ve marka bütünlüğü çalışmaları yapılmasına karar verildi.
    • Test senaryoları hazırlanarak kullanıcı deneyimi odaklı geliştirmeler yapılmasına karar verildi.
    • İngilizce veri ve PDF dokümanları toplanmasına karar verildi.
English Explanation
  • Sprint Notes:

    • A user interface was developed to receive natural language questions using Streamlit.
    • Google Embedding API was used to generate embeddings from user question sentences.
    • FAISS vector search algorithm was used to find the most relevant content.
    • Gemini-Pro LLM was integrated to generate responses from the most relevant content.
    • Figma was used for application interface design.
    • An input form was developed to collect user information such as age and gender.
    • Multilingual interface support was added to provide English language support.
    • A short demo video was created for application presentation.
    • Sprint planning and daily Scrum meetings were held regularly.
    • Task distribution and team tracking were managed through Trello.
  • Estimated story points to be completed in this sprint: 33 Points

  • Story point allocation logic: The project backlog consists of approximately 100 points in total.

    • Sprint 1 focused on Knowledge Base creation and completed 35 points,
    • Sprint 2 focused on Question-Answer Module and completed 33 points,
    • Sprint 3 will focus on UI, Deployment, and Testing, targeting 25–35 points.
  • Daily Scrum: Due to time constraints, daily Scrum meetings were conducted via WhatsApp.An example ODT file of the daily Scrum notes is shared in the README:📄 Sprint 2 Daily Scrum Notes

  • Product Backlog URL:Trello Board – Sprint 2

  • Sprint Review:

    During Sprint 2, we focused on implementing the Question-Answer module, one of the core components of our project.
    The main goal was to enable the system to generate meaningful and contextually accurate responses** to user questions in natural language.
    Key achievements in this sprint include:

    • Question input and embedding generation:User questions submitted via the interface were converted into vector embeddings using Google Text Embedding API.
      This enabled the system to retrieve the most semantically relevant content.

    • Vector indexing and search:Content data was indexed using FAISS to allow efficient similarity search in vector space.
      FAISS data was saved to disk and configured to be reloaded later when needed.

    • LLM integration:The most relevant content retrieved from FAISS was inserted into a prompt template and processed with Google Gemini-Pro.
      This generated unique and meaningful answers tailored to the user's question.

    • English language support:To reach a broader audience, an English interface option was added.
      Users can now choose their language at login and receive responses in the selected language.

    • User profile collection:The system was extended to collect basic user details such as age and gender.
      This lays the groundwork for future personalized responses.

    • Interface design:A modern, user-friendly interface was designed in Figma, balancing simplicity and functionality.

    • Demo video:A short video was prepared to showcase the application workflow and user experience.

    • Scrum process:Daily Scrum meetings were conducted regularly.
      The Scrum Master coordinated tasks and team progress, with task management handled via Trello.

Total 33 story points were planned and successfully completed. The system is now ready for the final sprint: UI, Deployment, and Testing.

  • Sprint Review Participants:Neslişah Ebral Durdu, Sıla Carus, Büşranur Delice, Ozan İnaltekin, Yavuz Selim Özdemir

  • Sprint Retrospective:

    • Decided to complete the application deployment process in the next sprint.
    • Planned to work on logo and brand consistency.
    • Test scenarios will be prepared to focus on user experience improvements**.
    • Additional English datasets and PDF documents will be collected.

Sprint 3

  • Sprint Notları:

    • Figma kullanılarak HeartHelper logo tasarımı yapıldı ve uygulama ile dokümanlara entegre edildi.

    • Streamlit Community Cloud üzerinde uygulama deploy edildi, gerekli requirements.txt ve secrets yapılandırmaları tamamlandı.

    • Google Gemini ve FAISS kullanılarak çok dilli (TR/EN) asistan soru-cevap yapısı oluşturuldu.

    • Kullanıcı profilleri (yaş, cinsiyet, hastalık) iki dilde hazırlanarak Streamlit arayüzüne eklendi.

    • Örnek kullanıcı profilleri ve sorular ile uygulama test edildi; yanıt doğruluğu, akış ve çok dilli destek kontrol edildi.

    • Kullanıcı deneyimini artırmak için ek sağlık önerileri ve yönlendirici bilgiler eklendi, küçük UX iyileştirmeleri yapıldı.

    • Yeni İngilizce PDF ve veri kaynakları toplanarak uygulamaya entegre edildi.

    • Sprint notları İngilizceye çevrilerek GitHub dokümanlarına eklendi.

    • GitHub repo güncellenerek kod açıklamaları ve README düzenlendi.

    • Scrum Master tarafından daily scrum, sprint review ve retrospective toplantıları yürütüldü; görev takibi Trello üzerinden sağlandı.

    • Sprint içinde tamamlanması tahmin edilen puan: 37 Puan

    • Puan tamamlama mantığı: Toplamda proje boyunca tamamlanması gereken ortalama 100 puanlık backlog bulunmaktadır. 1. sprintte Bilgi Tabanı Oluşturulacağı için 35 puanlık hedef tamamlanmıştır, 2. sprintte Soru-Cevap Modülü hazırlanıp 33 puan tamamlanmıştır, 3. sprintte Uygulama Arayüzü ve Yayına Hazırlık süreçleri ve test süreçleri hazırlanıp 37 puanlık hedef tamamlanmıştır.

  • Daily Scrum: Daily Scrum toplantılarının zamansal sebeplerden ötürü Whatsapp üzerinden yapılmasına karar verilmiştir. Daily Scrum toplantısı örneği odt olarak Readme'de tarafımızdan paylaşılmaktadır: 📄 Sprint 3 Toplantı Notları

  • Product Backlog URL: Trello Board - Sprint 3

  • Sprint board screenshotları:

image image image
  • Ürün Durumu: Ekran görüntüleri:
image image image
  • Ürün Tanıtım Videomuz: https://www.youtube.com/watch?v=TaYxlalz2f4

  • Sprint Review:

    • Figma kullanılarak HeartHelper markasına uygun, dikkat çekici ve anlamlı bir logo tasarlandı. Bu logo, hem uygulama arayüzüne hem de proje dokümanlarına entegre edilerek görsel bütünlük sağlandı.
    • Uygulama, Streamlit Community Cloud üzerinde başarıyla deploy edildi. Tüm dosyalar GitHub’da güncellendi ve requirements.txt ile secrets yapılandırmaları tamamlanarak uygulamanın sorunsuz şekilde erişilebilir olması sağlandı.
    • Google Gemini ve FAISS altyapısı kullanılarak çok dilli (Türkçe/İngilizce) asistan soru-cevap modülü geliştirildi. Bu modül, kullanıcıların doğal dilde sordukları sorulara, veri tabanındaki en alakalı içerik üzerinden yanıt üretebilecek duruma getirildi.
    • Kullanıcı profilleri (yaş, cinsiyet, hastalık türü) hem Türkçe hem de İngilizce olarak hazırlandı ve Streamlit arayüzüne entegre edildi. Bu sayede asistan, kullanıcıya özel ve kişiselleştirilmiş yanıtlar verebilir hâle geldi.
    • Uygulama, hazırlanan örnek kullanıcı profilleri ve sorularla test edildi. Yanıtların doğruluğu, akışın sorunsuz ilerlemesi ve çok dilli desteğin kararlılığı kontrol edildi. Tespit edilen küçük hatalar sprint içinde düzeltilerek uygulama stabilize edildi.
    • Kullanıcı deneyimini geliştirmek amacıyla ek sağlık önerileri, yönlendirici bilgiler ve içerik iyileştirmeleri uygulamaya eklendi. Ayrıca, arayüz üzerinde küçük UX geliştirmeleri yapılarak kullanım kolaylığı artırıldı.
    • Daha fazla İngilizce PDF ve veri kaynağı toplanarak uygulamanın içerik çeşitliliği artırıldı. Bu sayede, İngilizce dil desteği daha zengin ve anlamlı hâle getirildi.
    • Sprint notları İngilizceye çevrilerek GitHub’da paylaşıldı. Kod açıklamaları ve README dosyası güncellenerek proje dokümantasyonu son hâline getirildi.
    • Scrum Master tarafından günlük Scrum toplantıları, sprint review ve retrospective süreçleri etkin şekilde yönetildi. Takım koordinasyonu sağlandı, görev dağılımı ve ilerleme Trello üzerinden takip edilerek proje organizasyonu optimize edildi.

Bu sprintin sonunda HeartHelper uygulaması:

  • Çok dilli (Türkçe ve İngilizce) asistan soru-cevap modülüne sahip,

  • Kullanıcı profilleri ve ek sağlık önerileri ile desteklenmiş,

  • Streamlit Community Cloud üzerinde sorunsuz çalışan,

  • Güncellenmiş dokümantasyon ve GitHub reposu ile paylaşılabilir bir hâle getirilmiştir.

  • Sprint Review katılımcıları: Neslişah Ebral Durdu,Sıla Carus,Büşranur Delice,Ozan İnaltekin,Yavuz Selim Özdemir

  • Sprint Retrospective:

    Projemizin son sprintini tamamlamış bulunuyoruz ve HeartHelper uygulaması tüm planlanan özellikleri ile başarıyla geliştirilmiştir.

    • Takım içi iletişim ve iş birliği etkin şekilde sağlandı.
    • Scrum süreçleri (daily scrum, sprint review, retrospective) disiplinli bir şekilde yürütüldü.
    • Görevler Trello üzerinden düzenli takip edilerek iş akışı şeffaf ve planlı ilerledi.
    • Çok dilli destek ve kullanıcı deneyimi odaklı geliştirmeler başarıyla tamamlandı.
    • Test süreçleri düzenli yapılarak uygulamanın stabil ve hatasız çalışması sağlandı.
    • Bazı teknik entegrasyonlarda zaman yönetimi daha iyi planlanabilir.
    • Dokümantasyon ve notların erken sprintlerde de güncellenmesi faydalı olacaktır.
    • Kaynak ve veri toplama süreçleri daha erken başlatılabilir.

Genel olarak, ekip olarak proje boyunca gösterilen özveri ve iş birliği sayesinde hedeflerimiz zamanında ve yüksek kalitede gerçekleşti. HeartHelper projesi başarıyla tamamlanmış ve sunulmaya hazır durumdadır.

English Explanation
  • Sprint Notes:

    • HeartHelper logo was designed using Figma and integrated into the application and project documents.

    • The application was deployed on Streamlit Community Cloud, with all necessary requirements.txt and secrets configurations completed.

    • A multilingual (TR/EN) assistant question-answer structure was created using Google Gemini and FAISS.

    • User profiles (age, gender, medical conditions) were prepared in two languages and integrated into the Streamlit interface.

    • The application was tested with sample user profiles and questions; answer accuracy, workflow, and multilingual support were verified.

    • Additional health tips, guiding information, and minor UX improvements were added to enhance user experience.

    • New English PDFs and data sources were collected and integrated into the application.

    • Sprint notes were translated into English and added to GitHub documentation.

    • GitHub repository was updated with code explanations and README improvements.

    • Daily scrum, sprint review, and retrospective meetings were conducted by the Scrum Master; task tracking was maintained via Trello.

    • Estimated story points completed in this sprint: 37 Points

    • Story point allocation logic: The total project backlog consisted of approximately 100 story points.
      Sprint 1 focused on Knowledge Base creation and completed 35 points.
      Sprint 2 developed the Question-Answer Module and completed 33 points.
      Sprint 3 completed the UI, Deployment, and Testing processes, reaching 37 points.

  • Daily Scrum: Daily Scrum meetings were conducted via WhatsApp due to time constraints.
    An example ODT file of the daily Scrum notes is shared in the README: 📄 Sprint 3 Daily Scrum Notes

  • Product Backlog URL: Trello Board - Sprint 3

  • Sprint Review:

    • A logo suitable for the HeartHelper brand was designed in Figma. This logo was integrated into both the application interface and project documents to ensure visual consistency.
    • The application was successfully deployed on Streamlit Community Cloud. All files were updated on GitHub, and requirements.txt and secrets configurations were completed to ensure smooth accessibility.
    • A multilingual (Turkish/English) assistant question-answer module was developed using Google Gemini and FAISS. This module can generate meaningful and contextually relevant responses to users’ natural language questions.
    • User profiles (age, gender, medical condition) were prepared in both Turkish and English and integrated into the Streamlit interface. This allowed the assistant to provide more personalized answers.
    • The application was tested with sample user profiles and questions. Answer accuracy, workflow, and multilingual stability were verified. Minor issues detected during testing were fixed within the sprint to stabilize the application.
    • To improve user experience, additional health tips, guiding information, and content enhancements were added. Minor UX improvements were implemented to increase usability.
    • Additional English PDFs and data sources were collected to increase content variety, improving the richness of English language support.
    • Sprint notes were translated into English and shared on GitHub. Code explanations and the README file were updated to finalize the project documentation.
    • Daily scrum, sprint review, and retrospective processes were effectively managed by the Scrum Master. Team coordination was maintained, and task tracking was organized via Trello.

    By the end of this sprint, the HeartHelper application is:

    • Equipped with a multilingual (Turkish & English) Q&A assistant module,
    • Enhanced with user profiles and additional health recommendations,
    • Fully deployed on Streamlit Community Cloud,
    • Documented and shareable via GitHub.
  • Sprint Review participants: Neslişah Ebral Durdu, Sıla Carus, Büşranur Delice, Ozan İnaltekin, Yavuz Selim Özdemir

  • Sprint Retrospective:

    We have successfully completed the final sprint, and the HeartHelper application has been developed with all planned features.

    • Team communication and collaboration were effectively maintained.
    • Scrum processes (daily scrum, sprint review, retrospective) were conducted with discipline.
    • Tasks were tracked via Trello, ensuring a transparent and organized workflow.
    • Multilingual support and user experience-focused improvements were successfully implemented.
    • Testing was conducted regularly, ensuring the application runs stably and without errors.
    • Some technical integrations could benefit from better time management.
    • Documentation and notes would be more useful if updated in earlier sprints.
    • Data and resource collection processes could start earlier in future projects.

Overall, the dedication and teamwork of the team ensured that our goals were achieved on time and with high quality.
The HeartHelper project has been successfully completed and is ready for presentation.

About

Google Yapay Zeka ve Teknoloji Akademisi-Mezuniyet Bootcamp/Grup 7

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages