Inspiration
The idea for this project came from my own experience searching for certification deals online. I often had to check multiple sites, collect links, and spend time figuring out where everything was stored. This made me think of building a one-stop shop for certification offers and learning resources. I also wanted to include a chatbot that could instantly surface the right resources, without the hassle of digging through folders or bookmarks.
What it does
The project has two parts: Certification Coupon Hunter – A web application where the user enters their name, the certification they’re planning to take, and the provider. When they click Find Deals, the system fetches and displays relevant offers or posts for that certification using Google Search API. Learning Resource Chatbot – A conversational agent that provides learning resources based on the selected provider. I added quick-action buttons (e.g., “Find resources for AWS”) that return links to Skill Builder, workshops, and official documentation right away.
How I built it
The deal finder uses AWS Lambda as the core function. When a request is submitted, API Gateway triggers the Lambda, which calls the Google Custom Search API. The results are stored in DynamoDB in real time, and then retrieved and shown to the user. The chatbot is powered by a Bedrock agent integrated with Lambda. When a resource button is pressed, the agent invokes a Lambda action (learning_resources) to fetch curated links from DynamoDB and respond through the app. If the user asks free-form questions, the agent calls the Google Search API through the same Lambda to return results dynamically.
Challenges I ran into
Building this project came with several technical and architectural challenges: API Limitations – Many APIs I tested either required paid plans or had very low free request limits. Finding and configuring the Google Custom Search API with 1,000 free daily requests was a breakthrough, but it required handling quota errors and designing retry logic. Lambda Packaging & Dependencies – Managing Python dependencies in AWS Lambda was tricky. Some libraries required custom layers, and debugging cold start delays added complexity. State Management in DynamoDB – Storing and retrieving search results in real time meant designing efficient DynamoDB partition keys and indexes. At first, queries were slow or returned incomplete data until I optimized schema design. Bedrock Agent Integration – Getting the Bedrock agent to trigger Lambda actions consistently required fine-tuning action groups. Misconfigured intents often led to wrong actions being called. Chatbot Experience – Creating a smooth chat flow was harder than expected. Handling unexpected user input (outside of button clicks) meant building robust fallback logic so the agent wouldn’t break. Error Handling & Logging – Debugging serverless apps without a centralized log was time-consuming. I had to rely on CloudWatch logs and add structured logging to trace issues.
Accomplishments that I am proud of
I am proud to have built this project end-to-end, taking full ownership from idea to working prototype. This was my first time integrating Bedrock Agents with AWS serverless services like Lambda, API Gateway, and DynamoDB, and I successfully delivered a scalable, multi-provider solution covering AWS, Azure, GCP, Databricks, and Salesforce. Along the way, I overcame challenges with API limits, dependency management, and real-time DynamoDB queries, which helped me gain valuable skills in schema design, structured logging, and serverless debugging. Beyond the technical implementation, I focused on creating an intuitive user experience by combining deal discovery with a chatbot that delivers resources instantly through quick-action buttons or natural queries. Despite the hackathon’s time constraints, I delivered a polished prototype that demonstrates not just what’s possible now, but also a clear vision for evolving into a comprehensive certification and resource hub..
What I learned
There’s no single way to solve a problem—experimentation with open resources and cloud services is key. I learned to be more resourceful and resilient while troubleshooting.
What's next for Certification Resources & Deals Hunter
Next, I plan to expand the platform into a comprehensive certification hub that includes the latest news, advanced chatbot features, and more provider integrations—making it a go-to resource for professionals looking to upskill. I will plan to enhance this into a personalized certification hub that includes AI-driven recommendations, chatbot memory, and multi-language support.
Log in or sign up for Devpost to join the conversation.