Overview

On December 7th, 2024, the Greek Campus in Tahrir buzzed with excitement as Google DevFest Cairo 2024 brought together tech enthusiasts, developers, and innovators for a full day of inspiration and innovation. Organized by Google Developer Group (GDG) Cairo and supported by sponsors like Daytona, Sofco, Nsave, and Women Techmakers Cairo, this year's event was the largest yet, with 10,000 registrations—double last year's attendance. Packed with over 25 talks, hands-on workshops, and even lighthearted activities like pool, darts, and selfie stations, DevFest Cairo proved to be a vibrant celebration of technology, learning, and community.

Main Stage Wrap

Keynote

Mohamed Nagy, the head of GDG Cairo, kicked off the event with an inspiring keynote. He warmly welcomed the attendees, outlined the exciting plan for the day, and shared impressive facts about this year's DevFest, including its record-breaking 10,000 registrations. A video recap of last year's event added a nostalgic touch, setting the tone for an unforgettable day. Nagy concluded with a heartfelt wish for everyone to enjoy the event and make the most of the incredible lineup of talks, workshops, and activities.

Models, Platforms & Datasets by A. Mahdy Abdelaziz

In a thought-provoking session, A. Mahdy Abdelaziz broke down the intricate process of building AI models, from collecting and cleaning data to preparing training and testing datasets. He highlighted how iterative and resource-intensive this process can be, making it ideal only for businesses whose core focus is model development. For others looking to integrate AI capabilities, he advised leveraging existing models and fine-tuning them for their specific use cases instead of reinventing the wheel. Abdelaziz wrapped up with a crucial takeaway: decide whether your business is about building models or using them—this choice defines your strategy moving forward.

A Framework to Build Human-First AI Applications by Asmaa Ibrahim

In her engaging talk, Asmaa Ibrahim outlined a framework for creating AI-powered applications with a focus on Generative AI (GenAI). She emphasized that everyone can play a part in the GenAI ecosystem by identifying their role, which she broke down into four key categories:

1. Technology Providers
   The builders of foundational AI models, focusing on scalability, efficiency, and ethics. Providers create the "digital library" of tools and capabilities that enable businesses to innovate.

2. Technology Brokers
   The connectors who simplify access to GenAI by creating tools and platforms like LangChain, bridging the gap between complex AI technology and real-world applications.

3. Technology Consumers
   The creators of AI-powered solutions. Consumers use GenAI tools to solve specific challenges, like developing chatbots or predicting trends, and turn foundational models into scalable business applications.

4. End Users
   The beneficiaries of GenAI-powered tools. From language translation apps to financial management platforms, end users experience the practical benefits while providing feedback to refine these solutions.

Ibrahim encouraged the audience to reflect on which role suits them best, noting that every role is essential to unlocking the full potential of GenAI. For a deeper dive, she pointed to her Medium article, "GenAI's Ecosystem: Discover Your Role".

FinOps: A New Hope by Seifeddin Mansri

In his insightful talk, Seifeddin Mansri discussed the growing dependence on cloud services and the accompanying rise in costs associated with running or using them. To address these challenges, he introduced FinOps as a new hope, drawing an analogy to Star Wars: A New Hope.

At its core, FinOps is a cultural practice—a framework that enables teams to manage cloud costs effectively. It fosters a culture where everyone takes ownership of their cloud usage, supported by a central best-practices group. Cross-functional teams spanning Engineering, Finance, Product, and other departments collaborate to:

• Enable faster product delivery
• Gain more financial control
• Achieve better cost predictability

The term FinOps combines “Finance” and “DevOps,” emphasizing communication and collaboration between business and engineering teams.

Mansri also showcased FOCUS, an open-source specification that defines clear requirements for cloud vendors to deliver consistent and reliable cost and usage datasets. This tool aims to streamline cloud cost management for organizations.

To learn more about FOCUS, visit FOCUS official website.
For more information on FinOps, check out the FinOps Toolkit.

Demystifying OAuth2 & OIDC by Nejmeddine BEN OUARRED

Nejmeddine BEN OUARRED delivered a fascinating talk where he demystified the architecture of OAuth2 and OIDC (OpenID Connect), using an analogy to the Cairo Metro System to make the concepts engaging and easier to understand.

He broke down how each system works:

1. OAuth2
   OAuth2 acts like a ticketing system in the metro. Just as you need a valid ticket to pass through the gates and access the metro system, applications need an access token to access resources securely. It ensures that only authorized individuals (or applications) can use specific services.

2. OIDC
   OpenID Connect, built on top of OAuth2, is akin to a metro card with your photo and personal details. It not only allows you access to the metro but also confirms your identity. OIDC adds an authentication layer, enabling systems to verify who you are while using the token for secure access.

Through this analogy, Nejmeddine highlighted how OAuth2 focuses on authorization (what you can do), while OIDC adds authentication (who you are).

His creative approach made these technical concepts relatable and memorable for the audience.

From an Engineer to a Founder by Abdalah AbuHashem, nSave Founder

Abdalah AbuHashem shared his journey and insights in his talk, "From an Engineer to a Founder," where he explored what it takes to transition from being an engineer to founding a startup. He emphasized that while being a skilled engineer is crucial, it’s only part of the equation.

Key takeaways from his talk included:

1. Engineering Skills Are Essential, but Not Sufficient
   Being an excellent engineer is a strong foundation, but building a successful startup requires much more than technical expertise.

2. Understand the Problem You're Solving
   To succeed as a founder, you must deeply understand the problem you're addressing. This comes from actively talking to users, learning their pain points, and staying open to insights that might shift your perspective.

3. Balancing Skills with the Right Tools
   AbuHashem encouraged founders not to stress too much about choosing the "perfect" tool for the job. Instead, he suggested combining:

   • The skills you excel at (where you already have strong developer experience or DX).
   • The right tool for the job, ensuring a balance between expertise and effectiveness.

His talk highlighted the importance of adaptability, user-centric thinking, and leveraging both technical and non-technical skills to succeed as a founder.

Building the Future: Software Engineering Skills Before and After AI by Islam Elgamal

In his thought-provoking talk, Islam Elgamal explored how software engineering has evolved with the advent of AI, comparing the tools and methodologies used before AI and after AI.

Before AI

• Engineering tools and practices were more reliant on waterfall models, where development followed a linear and structured process.
• There was less automation, leading to more manual effort in repetitive tasks.

After AI

• The shift towards agile frameworks has enabled faster and more iterative development cycles.
• The incorporation of AI-powered tools has automated numerous aspects of software engineering, from code generation to testing and deployment, boosting efficiency and productivity.

Key Message

Elgamal encouraged young learners to adopt a mindset of continuous learning, emphasizing that adaptability and resilience are now essential skills in the ever-changing landscape of software engineering. He stressed that as AI continues to reshape the field, professionals must remain open to change, continuously updating their skills and embracing new tools and methodologies.

His talk served as both a reflection on the transformation of the industry and a call to action for engineers to prepare for a future driven by innovation.

RAGs to Riches: Generating Sustainable Travel Recommendations with RAG and Gemini by Ashmi Banerjee

Ashmi Banerjee shared insights from her team's project focused on creating sustainable travel recommendations using Retrieval-Augmented Generation (RAG). The talk highlighted their methodology, the challenges they faced, and their key findings in leveraging large language models for personalized and sustainable travel solutions.

Key Highlights

1. Exploration of Models

   • The team experimented with multiple large language models (LLMs), including Llama, Mistral, and Gemini.
   • They used these models as judges to evaluate and refine the recommendations, ensuring contextual relevance and sustainability.

2. Data Source

   • The project utilized data from WikiVoyage, a reputable source of travel-related information.
   • The knowledge base was enriched with sustainability metrics, including eco-friendly accommodations, carbon-neutral transportation options, and environmentally responsible attractions.

3. Challenges in Consistency

   • While one model outperformed others in certain aspects, the team found that the results were inconsistent, particularly regarding fairness and overall reliability.
   • This underscored the complexity of balancing performance with ethical considerations in generating travel recommendations.

4. RAG and Sustainability

   • The team adapted RAG to the travel domain, focusing on generating recommendations tailored to diverse user preferences while promoting environmentally responsible travel practices.

Expected Outcomes and Future Directions

• A robust recommendation system providing personalized, sustainable travel suggestions for European cities.
• Insights into the performance of different LLMs in generating contextually relevant and sustainable recommendations.
• Recommendations for future optimizations to enhance the system's usability and promote sustainable travel.

Conclusion

Banerjee emphasized the potential of combining advanced NLP techniques and sustainability-focused data to create impactful solutions for the travel domain. Her talk served as an example of the challenges and opportunities in leveraging AI for ethical and practical applications.

To learn more about the project, explore Gemma on Hugging Face and the RAG methodology.

A New Day of AI: What's New in Azure AI by Ahmed Bahaa

Dr. Ahmed Bahaa, Microsoft Regional Director for Egypt and a globally recognized thought leader in IT, delivered a visionary and thought-provoking talk on the future of AI, quantum computing, and societal change. With 29+ years of experience, 45+ certifications, and numerous accolades, Dr. Bahaa combined deep technical insights with bold predictions about technology and its impact on society.

He began his talk by paying tribute to the resilience of Palestinians, honoring their sacrifices and tying their struggle for freedom to the global need for resilience and unity in overcoming challenges. Dr. Bahaa also recognized the foundational contributions of John J. Hopfield and Geoffrey Hinton to artificial neural networks, whose pioneering work has shaped the AI landscape as we know it today.

Dr. Bahaa painted a bold future for AI, envisioning a world where interconnected agents, independent AI systems, collaborate seamlessly to solve problems with minimal human intervention. These agents could reduce the need for human oversight, creating an ecosystem where humans step in only when AI fails, enabling more efficient and innovative solutions.

He shared striking insights into the rapid advancements in quantum computing, predicting that retail quantum computers might soon become available for home use. This development, he noted, could revolutionize computational capabilities, marking a significant milestone in technological innovation. However, he also warned of the profound implications for security, highlighting that by 2035, the National Institute of Standards and Technology (NIST) may actively prohibit the use of 128-bit encryption algorithms, as they will no longer be sufficient in a quantum-powered world.

Dr. Bahaa revisited a 2016 prediction that 65% of children starting school that year would eventually work in jobs that didn’t exist at the time. To illustrate this, he highlighted the rise of roles like prompt engineers, a position that has become vital in the AI-driven world but was unheard of just a few years ago. This prediction underscores the rapid and unpredictable evolution of the job market and its demand for new skills.

He also remarked on the acceleration of Moore’s Law, noting that while it held true until 2022 with technological advancements doubling annually, progress has since accelerated to a pace where innovations occur every six months. Among his more provocative predictions, he speculated about a future where robots and humans might rise together to demand societal recognition for human-robot relationships, reflecting the ethical dilemmas that may emerge as AI continues to integrate into everyday life.

Dr. Bahaa concluded his talk by urging young engineers and students to embrace adaptability and resilience as essential traits for navigating a world of constant change. He reminded the audience that change is the only constant, and those unwilling to adapt will struggle to thrive in the rapidly evolving technological landscape.

Blending his deep expertise with bold and provocative ideas, Dr. Ahmed Bahaa delivered a talk that left the audience inspired, reflective, and eager to prepare for a future shaped by AI, quantum computing, and societal transformation.

Secondary Stage Wrap

Prompt Engineering: Strategies for Better Results by Islam Mahmoud

Islam Mahmoud delivered a practical and insightful session on crafting better prompts to achieve faster and more accurate results when working with language models. He shared several effective strategies:

1. Adding a Persona: By specifying the context or the audience, prompts can produce highly tailored outputs. For instance:

Each prompt yields different results based on the specified persona and audience.

2. Breaking Tasks into Steps: For complex tasks, Mahmoud recommended breaking them down into smaller steps. For example, if you're processing audio data:

   • Provide the language model with a step-by-step guide, such as extracting the transcript first, followed by text processing.

3. Iterative Refinement: He emphasized re-iterating prompts to focus on specific aspects of the task. For instance:

Re-iterating helps to refine the response and align it with the exact requirements.

4. Adjusting Model Parameters: Mahmoud discussed using temperature values to control creativity:

   • 0 for deterministic, consistent results.
   • 1 for creative outputs.
     He also highlighted limiting response length (number of tokens) to reduce costs, especially when using paid models. For example, asking Gemini to write a one-sentence definition for DevOps can save both time and money.

5. Providing Schemas for Structured Outputs: To ensure concise and relevant responses, he recommended defining output schemas. As an example, he shared how he built an app to extract only the necessary information from real estate agent messages for him and his wife to compare properties. A structured prompt might look like:

Mahmoud's talk was a treasure trove of practical tips for maximizing the potential of language models while optimizing efficiency and cost.

Refactoring Guru by Eslam Medhat

Eslam Medhat delivered an insightful talk on the importance of refactoring—the process of improving code without changing its functionality. Drawing from the popular book Refactoring Guru, he outlined essential techniques to clean up messy code and shared tips on identifying the right time to refactor. Medhat emphasized that refactoring isn't just about aesthetics; it enhances code readability, maintainability, and long-term performance. His talk provided an accessible overview of best practices for developers looking to write cleaner, more efficient software.

The Full Experience

DevFest Cairo 2024 wasn't just about tech—it was about fostering connections and creating memories. Between talks and workshops, attendees enjoyed fun activities like pool, darts, and selfie stations, adding a touch of lightheartedness to the day. These moments of relaxation created a sense of community and camaraderie that set DevFest apart from typical conferences.

Reflections on the Event

DevFest Cairo 2024 wasn't just an event—it was a movement. From awe-inspiring talks like Ahmed Bahaa's futuristic vision to hands-on workshops that bridged the gap between theory and practice, the day was a celebration of innovation, creativity, and collaboration. The staggering 10,000 attendees and electric energy underscored Cairo's growing prominence as a hub for technological excellence.

As I left the Greek Campus, I couldn't help but feel inspired. DevFest Cairo didn't just offer a glimpse into the future of technology—it empowered us to shape it. If you want to take a glimpse of the event, check out the pages of Google Developer Groups Cairo on social media platforms like LinkedIn, X.com, or their official website. Here's to even bigger ideas and bolder conversations at DevFest 2025!

---

Note: While I've shared highlights from the sessions I attended, I couldn't cover every talk or workshop due to the packed schedule. I apologize for any missing details and encourage you to explore GDG Cairo’s channels to learn more about the full event.