How Are We Making Technical and Business Decisions to Build Our Video Ad Platform

I'm currently participating in the world's largest hackathon, and I want to share the behind-the-scenes journey of what we're building, how we're developing it, and the reasoning behind every critical decision we've made. This is a real-time case study of product development, technical architecture choices, and startup pivots.

The Current State: What We're Building

Currently, we are developing a video ad platform that processes campaigns through twelve distinct stages. The application uses a hybrid development approach in which I started with Bolt.new for rapid prototyping, then switched to VS Code for more complex backend development.

Our Tech Stack:

• Frontend: React
• Backend: Python with FastAPI
• Architecture: Separated frontend and backend for scalability
• Deployment: AWS Lightsail instances
• Integration: GitHub for version control with automated deployments

The reason for this hybrid approach? Bolt.new doesn't support Python coding yet, and I had specific reasons for choosing Python for the backend (which I'll explain below).

Sharing the Journey: Documentation and Community

So far, I've shared nine videos and multiple posts on X and LinkedIn documenting this hackathon experience. My goal is to help other participants by sharing knowledge, ideas, and practical insights about:

• How to participate effectively in hackathons
• Creating and managing Bolt.new accounts
• Finding and validating ideas worth building
• Technical implementation strategies

I've also created a comprehensive AI coding tool resource that I'm continuously updating with new content. This documentation serves both as a learning resource for others and a accountability mechanism for my own progress.

The Great Pivot: From Shopify SEO to Video Ads

The Original Idea (May 29 - June 8)

When the hackathon started on May 29th, my co-founder and I initially decided to build a Shopify SEO autopilot tool. We spent an entire week in research mode:

Market Research Process:

1. Competitor Analysis: Studied 3-5 existing products in the space
2. Feature Comparison: Analyzed their capabilities and offerings
3. Customer Feedback: Read reviews (both positive and negative)
4. Video Content Review: Watched their promotional and tutorial content
5. Customer Discovery: Reached out to Shopify store owners directly
6. Pricing Research: Understood willingness to pay for solutions

Technical Planning:

• Created work breakdown structures
• Designed application flow diagrams
• Mapped out technical requirements

Why We Pivoted

After a week of intensive research, two critical issues emerged:

1. Market Saturation: The Shopify SEO space is incredibly crowded with established competitors
2. Late Market Entry: We would be entering as very late comers to a mature market

The New Opportunity

During our research phase, my co-founder connected with Victor, who has been running an ad agency since 2019. After several meetings to understand each other's backgrounds and expertise, we identified an opportunity to convert his manual agency processes into a scalable product.

Due Diligence Process:

• Multiple discovery sessions to understand manual workflows
• Competitive analysis in the video ad space
• Process mapping and optimization opportunities
• Technical feasibility assessment

On June 8th, we made the decision to pivot to the video ad platform we're now building.

Technical Architecture Decisions

The Four Proof of Concepts

Before committing to our final architecture, I built four different proof of concepts to validate technical approaches:

POC 1: Next.js Full-Stack

• Stack: Next.js with API routes
• Integration: Multiple APIs and database connections
• Outcome: Worked well but concerned about frontend security and performance at scale

POC 2: React + Node.js/Express

• Stack: React frontend, Node.js/Express backend
• Architecture: Proper separation of concerns
• Outcome: Clean architecture but staying within JavaScript ecosystem

POC 3: React + Python/FastAPI

• Stack: React frontend, Python/FastAPI backend
• Architecture: Mixed technology stack
• Outcome: Leverages Python's AI/ML strengths

POC 4: React + Python/Django

• Stack: React frontend, Django backend
• Architecture: Full-featured Python framework
• Outcome: Too heavy for our focused API needs

The Final Decision: React + Python/FastAPI

After extensive research, consulting with other developers, and testing with various LLMs, I chose the React + Python/FastAPI combination.

Advantages:

• AI/ML Optimization: Python excels in AI-related tasks
• Code Efficiency: Python allows more functionality with less code
• Performance: Separated architecture enables independent scaling
• Security: Backend keys and sensitive logic isolated from frontend

Trade-offs:

• Team Complexity: Requires knowledge of two technology stacks
• Hiring Challenges: Need developers skilled in both or separate teams

Alternative Consideration: The React + Node.js approach would have allowed training developers on a single technology stack (JavaScript), which offers advantages for scaling and team management.

Deployment and Infrastructure Decisions

Platform Evaluation

I considered multiple deployment options:

Platforms Evaluated:

• Heroku: Familiar but expensive at scale
• Railway: Good developer experience
• AWS EC2: Full control but complex setup
• AWS Lightsail: Simplified AWS experience
• Vercel: Great for frontend, limited Python support

Why AWS Lightsail?

I ultimately chose AWS Lightsail based on several factors:

1. Expert Recommendation: My friend Ush (13 years cloud experience, YouTube channel) recommended Lightsail
2. Scalability: AWS won't limit growth when we need to scale
3. Startup Credits: Received $1,000 in AWS startup credits
4. Cost Predictability: Fixed pricing model vs. variable bandwidth costs
5. Learning Opportunity: Wanted hands-on AWS experience

Current Infrastructure:

• Instance 1: Linux + Node.js (Frontend)
• Instance 2: Linux + Python/FastAPI (Backend + Database)

Automated Deployment Pipeline

I spent five hours setting up automated deployment from GitHub to AWS instances. The workflow:

1. Development: Code in Visual Studio Code
2. Version Control: Push to GitHub
3. Automated Deployment: GitHub Actions deploy to AWS instances
4. Testing: Validate functionality on live instances

This setup ensures consistent deployments and reduces manual deployment errors.

Project Management and Documentation

Current Work Breakdown Structure

Our development is organized into these key areas:

1. User Registration & Onboarding Flow
2. User Profile Management
3. Prompt Engineering (for AI components)
4. Database Design & Implementation
5. Application Architecture
6. Frontend Development
7. Backend API Development

Design and Planning Artifacts

We've created comprehensive documentation:

• Wireframes: Visual layouts for all major screens
• High-Level Flow: Overall user journey mapping
• Low-Level Flow: Detailed process workflows
• Technical Architecture: System design diagrams

Timeline and Next Steps

Current Status (June 17, 2025)

We're about halfway through the hackathon timeline. While the research and architecture phases took longer than initially planned, we have a solid foundation and clear path forward.

Upcoming Milestones

With approximately 15 days remaining, our goals are:

1. Complete Development: Finish core application features
2. Closed Beta Launch: Deploy to interested beta users
3. User Feedback: Gather insights from agency owners and business owners
4. Iteration: Refine based on real user feedback

We already have several business owners and agency owners who have expressed interest in testing the application, which validates our market approach.

Lessons Learned So Far

1. Research Prevents Costly Mistakes

Our week-long research phase for the Shopify SEO idea saved us from entering an oversaturated market. Thorough research is an investment, not a delay.

2. Technical Proof of Concepts Are Essential

Building four different POCs helped us make an informed technical decision rather than guessing. The time invested in testing approaches paid off in confidence and clarity.

3. Infrastructure Planning Matters

Setting up proper deployment pipelines early saves time and headaches later. Automated deployments are worth the initial setup investment.

4. Documentation Drives Clarity

Creating detailed work breakdown structures and flow diagrams helps maintain focus and provides clear development direction.

5. Community and Mentorship Accelerate Progress

Getting advice from experienced developers (like my friend with 13 years of cloud experience) prevented costly mistakes and accelerated decision-making.

Key Takeaways for Hackathon Participants

If you're participating in hackathons or building your own products, here are my recommendations:

Before Building

1. Invest in Research: Spend time understanding the market and competition
2. Talk to Potential Users: Real customer feedback trumps assumptions
3. Plan Your Architecture: Build POCs to validate technical decisions
4. Document Everything: Clear documentation accelerates development

During Development

1. Set Up Automation Early: Deployment pipelines save time and reduce errors
2. Choose Technologies Wisely: Consider team skills, scalability, and market factors
3. Plan for Scale: Architecture decisions are hard to change later
4. Stay Connected: Share progress and get feedback from the community

Technical Considerations

1. Security First: Separate sensitive logic from frontend code
2. Performance Planning: Consider deployment and scaling early
3. Tool Selection: Choose tools that support your specific needs (like Python for AI)
4. Backup Plans: Have alternative approaches ready

Looking Forward

The next two weeks will be crucial as we move from architecture and planning into full development mode. We're excited about the potential impact of converting manual agency processes into a scalable platform.

I'll continue sharing updates about our progress, challenges, and solutions. If you're working on similar projects or participating in hackathons, I'm happy to share resources, provide feedback, or collaborate on solutions.

The hackathon community is incredibly supportive, and I believe in helping each other succeed. Whether you need technical advice, want to share ideas, or just want to connect with fellow builders, reach out.

What's your biggest challenge in hackathons or product development? Share your experience in the comments below.

---

Follow my hackathon journey and get access to all my development resources, wireframes, and technical documentation. Connect with me on X and LinkedIn for real-time updates.