Upskill Coach - Powered by AWS AI Agent

AI-powered career coach built on AWS, guiding global talent to meet evolving skill demands with smart career path support, certification, and resume tips. Also integrates with Organization's LMS.

Comment

Inspiration

🌐 In today’s fast-paced tech world, professionals often feel lost trying to figure out which skills, certifications, or career paths will keep them competitive. Traditional learning platforms offer limited personalization, and career coaching is either costly or hard to access.

💡 That’s where the idea for Upskill Coach was born — a vision to democratize career coach using autonomous AI agents.

🛠️ Built on AWS’s advanced AI services, Upskill Coach guides students, professionals, and job seekers through personalized upskilling journeys.

🏢 Recognizing that organizations also face challenges in consistently upskilling their workforce, the solution was designed to work independently or integrate seamlessly with existing Learning Management Systems (LMS), enabling scalable, AI-powered career development for entire teams

What it does

Upskill Coach is an AI-powered career coach designed to help individuals advance in AI and technology. It provides:

🚀 Career Guidance: Discover AI roles, career paths, and industry insights. 📚 Learning References: Courses, certifications, and study materials—with optional LMS integration for organizations. 💻 Technical Skills: Explore programming languages, frameworks, and tools. 📝 Professional Development: Get resume tips, interview prep, and networking advice.

It’s more than a chatbot—it’s an autonomous agent that reasons, retrieves, and acts to guide users toward career success, whether used independently or integrated into enterprise learning systems.

How we built it

🤖 We built Upskill Coach using AWS services to enable autonomous AI actions such as reasoning, retrieving content, maintaining memory, and executing tasks.

⚙️ The backend logic is powered by AWS Lambda, which interacts with Bedrock Agent to handle user queries intelligently.

🌐 The frontend is a lightweight HTML page hosted on Amazon S3 and served globally via CloudFront for fast and easy access.

🏗️ This architecture ensures a scalable, serverless, and low-maintenance solution that’s both developer-friendly and enterprise-ready.

Challenges we ran into

🔄 Lambda Origin Parameter Conflict: Initially, we defined the Origin parameter in both the user interface and the Lambda function code. This caused unexpected behavior and blocked responses from reaching the frontend. Using Amazon CloudWatch logs and Amazon Q for troubleshooting, we identified the conflict and resolved it by removing the Origin parameter from the Lambda code. This restored proper communication between the UI and backend.

🧩 Formatting Responses in index.html: The raw responses from the agent needed to be parsed and rendered cleanly in the chat interface. We encountered issues with the output. Using Amazon Q for troubleshooting, we fixed the formatting logic and successfully displayed the agent responses in the UI.

Accomplishments that we're proud of

🤖 Built a fully functional autonomous agent using AWS services and delivered a smooth, end-to-end demo experience for the hackathon. 🚀 Streaming-safe Bedrock Agent Lambda: We successfully deployed a robust AWS Lambda integration with Bedrock Agents, handling streaming responses with UTF-8 encoding and emoji-safe output. 🧠 Multi-turn memory with fallback logic: Our agent supports contextual multi-turn conversations with resilient fallback handling, enabling smoother user experiences and graceful recovery. 🎨 Emoji-rich chatbot UI with visual polish: Our chatbot interface was designed for public-facing demos, featuring emoji-safe rendering, branded styling, and seamless integration with agent responses.

What we learned

🧠 How to design and implement autonomous agents using AWS services like CloudFront, Route53, S3 Bucket, Lambda Function, AWS Bedrock Agents (Claude vs Nova), DynamoDB, IAM Roles & Policies, CloudWatch, and Billing & Pricing. We also explored AWS AgentCore, Lex, and Kendra.

🛠️ Best practices for integrating LLMs, prompt engineering, and tool configuration.

⚡ How to build scalable, serverless AI applications with rapid iteration. Also how to Use Amazon Q for Faster AI Agent Development in AWS

What's next for Upskill Coach - Powered by AWS AI Agent

🔍 Amazon Kendra (RAG) can be leveraged for organization databases and documents — enabling smarter, context-aware search and retrieval.

🔄 Continuous Improvement: Implement a feedback mechanism to refine agent performance, personalize recommendations, and enhance user experience.

Built With

Share this project:

Updates