Build Production-Ready AI Agents and Multi-Agent Systems from JHU

I recently built a project that started as a Text-to-SQL agent and evolved into a broader multi-source Agentic AI orchestration platform using LangGraph, Ollama, ChromaDB, FastAPI, and React. The goal was to let a business user ask a question in natural language and have the AI understand the intent, identify the right source, and return the answer. The platform can orchestrate across databases, policies, documents, web search, and media. The most interesting engineering challenge wasn’t connecting an LLM—it was designing how the system decides where to look for the answer. I built the orchestration around intent understanding, source identification, agent and tool selection, context retrieval, reasoning, validation, guardrails, and grounded responses. One of the key principles I followed was not letting the LLM control everything. The LLM handles reasoning and orchestration, while deterministic components handle critical operations like SQL validation, read-only execution, query controls, schema-aware retrieval, source attribution, and human confirmation where required. This creates a better balance between AI autonomy and enterprise control. The bigger vision is to build an AI system that can identify the right source and provide a grounded answer. This project has been a great hands-on exploration of Agentic AI, multi-source orchestration, RAG, Text-to-SQL, AI security and guardrails, and enterprise AI architecture, and I’m continuing to enhance it for real-world use cases.

Read More

Years ago, I was a software engineer, but after moving into other roles my coding skills faded. Going into the Certificate Program in Agentic AI from Johns Hopkins and Great Learning, I honestly wondered whether I would be fast enough to pick coding up again. My original goal was to get under the hood, understand how the engine works, and grasp the mechanics behind agentic workflows so I could guide others and support my teams with more clarity. But the program ended up shifting something bigger. I didn’t just gain the strategic perspective to architect systems and support teams with confidence—I also got inspired to step onto the field. Understanding Agentic AI systems made me want to join the builders, support them, and build agents myself. Now I have the hands-on capability to build systems, along with the strategic view to guide others. At first, the learning felt like a puzzle, but once I started putting concepts into practice, a clear picture emerged. One of the biggest lessons for me was understanding trade-offs—balancing cost, quality, speed, and security is where real system design lives. Inspired by the course, I even developed an interactive AI-driven interviewer tool for my company to help executive leaders uncover their AI vision and growth opportunities. If you’re considering the program, don’t let the technical jump intimidate you. Huge appreciation to the team for keeping the content sharp, the live masterclasses current, and the pace challenging without letting people fall behind.

Read More

I’m happy to share that my third and final project in the Johns Hopkins University Certificate Program in Agentic AI was approved, which means I have now completed the program curriculum. During the course, I applied freshly gained theoretical knowledge in three practical case projects, which are summarized in my e-portfolio. The course combines theory, the current state of the technology, and hands-on agent creation. The good content, clear structure, and strong knowledge delivery made this a very well-condensed learning package. I have now received the certification, and I would definitely recommend the JHU and Great Learning curriculum.

Read More

I’m a marketing and business professional with more than 25 years of experience, and my career has increasingly moved toward the intersection of marketing, technology, commercial strategy, and operational excellence. At this stage of my career, I didn’t want AI to be something I simply used without understanding what was happening underneath it. I wanted to understand how AI systems actually work, especially how agentic AI could be applied to real business problems. My goal wasn’t to become a software engineer, but to become a more capable business professional who understands what AI can do, what its limitations are, and how to translate that capability into business value. That has already happened. The biggest surprise has been how quickly the learning moved from the classroom into my everyday work. I now use AI agents and agentic approaches in my daily professional life to work faster, structure information better, explore ideas more deeply, and improve the quality of the final outcome. It changed the way I think about productivity. I no longer see AI simply as a tool that helps me do my existing job faster; I increasingly see it as something that can change how the job itself is designed. My favorite part was when the concepts stopped being theoretical and I had to build something myself. Working on projects like an AI financial analyst and a multi-agent mortgage underwriting system was a turning point, because it showed me I could look at AI through the lens of business problems, processes, risks, and outcomes. That was extremely empowering. The biggest challenge was stepping outside my comfort zone as a non-technical professional, but I stayed hands-on, asked questions, experimented, made mistakes, and kept going. My advice is simple: don’t wait until you feel ready for AI. Come with curiosity, be willing to get uncomfortable, and build things. After 26 years of experience, this programme reminded me that experience and continuous learning don’t compete with each other—they amplify each other. AI isn’t replacing my experience; it is giving me new ways to use it.

Read More

I’m a Senior Delivery Manager with over 22 years of IT experience, based in Bengaluru, and I recently stepped down from my role at CDW to pursue a structured, deep-dive learning journey in AI. Before joining this program, I had already led AI transformation initiatives, completed certifications, and spent countless hours learning from tutorials, but I still lacked the confidence that I truly understood what was happening behind the scenes. I joined because I wanted to know AI, not just use it, and to become a more transformational leader who could guide organizations with conviction on when AI is truly the right solution. My favorite part of the program has been the hands-on coding. Building AI agents through Python gave me back something I had missed since my early years as a developer, and the end-to-end projects brought the course learnings together in a way no tutorial ever could. Prompt engineering was my biggest gap coming in, but the pre-work and mentor demonstrations completely changed how I approached LLMs and showed me a much deeper way to work with them. This program is the best thing that has happened to me in my structured learning journey. In just 18 weeks, I went from a leader with a strong traditional software delivery background to a consultant who can confidently guide organizations on their AI transformation journey. This program is already opening doors and generating meaningful interview conversations for my next role. It doesn’t just help you build your first AI agent; it guides you toward making it trustworthy.

Read More

I’m an Enterprise Architecture leader based in Pune with 18 years in technology, spanning software engineering, architecture, and technology strategy. I joined this program for two reasons. First, AI removed a constraint I had lived with for years: it let me move ideas from concept to something tangible without depending entirely on development effort, budget, and multiple teams. I could build, test, and validate ideas myself. Second, because I lead AI governance globally, I wanted to understand where agentic systems actually break, what autonomy really costs, and which controls are meaningful rather than decorative. The hands-on work was my favourite part. Building an Autonomous Financial Analyst and a multi-agent Mortgage Underwriting System taught me as much about governance as it did about engineering, and it directly influenced how I think about policy-as-code. The biggest challenge was balancing an intensive program with a demanding leadership role, travel, family health, and being present for my children. What made it possible was the consistent support from Ehetsham, who responded quickly whenever I needed help, and the guidance from Bhaskerjeet, who connected concepts to real problems. My advice to anyone considering this program is simple: don’t just learn AI, build with it. Bring your own problems in, challenge what you’re taught, and leave with at least one idea turned into something that runs. At this stage of my career, one thing is clear: continuous learning isn’t optional, it’s part of the job.

Read More

I’ve completed the Certificate Program in Agentic AI, a field that changes every week. Yet the teachers and mentors were extremely knowledgeable and responsive, and the course has enabled me to confidently enter the market with new AI-driven services. What’s even better, the support team has been outstanding.

Read More

I took the Certificate Program in Agentic AI, and it was very enlightening. As an Engineering Lead looking to integrate AI-assisted development into my team, this course gave me solid building blocks to guide version 1 of the AI development platform we currently use. I recommend it.

Read More

The JHU Agentic AI program by Great Learning was a well-structured and insightful learning experience. It offered a good balance of conceptual understanding, practical examples, and hands-on projects. The faculty was knowledgeable and supportive throughout. Overall, the course provided valuable exposure to Agentic AI and its practical applications.

Read More

This is a serious engineering program that teaches learners to actually design, build, and deploy autonomous systems. The curriculum is dense and practical: you go from RAG to production security, observability, and building multi-agent systems with LangGraph and distributed MCP. There’s a strong focus on real-world problem-solving, and the masterclasses and weekly mentor sessions are an incredible value-add for learning cutting-edge, industry-validated tips and tricks. The JHU faculty lectures give you the why behind the architecture, and the Great Learning support keeps you on track. It’s a serious 18-week commitment at 8 to 10 hours a week, but if you want to gain the skills to design and ship autonomous AI systems, this is the place to do it.

Read More