How My Summer Internship Kickstarted My Tech Journey (and How I Thrived Through It!)
Chapter 1: The Unexpected Buzz
It was a typical lazy evening, and there I was—Diya, the author of this blog, knee-deep into some random Netflix series when my phone buzzed. You know that feeling, right? When you’re so immersed that any notification feels like an unwelcome intruder. But this time, it was my friend, and they said something that immediately got my attention: “Ever heard of Garje Marathi?” I figured I’d humor the idea and check it out—why not, right? Fast forward a few clicks, and BAM! My tech journey was kick-started in the blink of an eye. As a Computer Science major and Math minor, I knew that everything was going to be extremely technical, but little did I know I would make a family through this internship, and gain more personal skills on top of the high level phD research that we ended up getting into.
Meanwhile, in Kunj’s world, things started a bit differently. He was added to a WhatsApp group—a scenario we all know too well. Usually, it’s just a barrage of irrelevant texts. But this time, something was intriguing. "Siddharth casually introduced us to the Garje Marathi community," he recalls. "One minute I’m mindlessly scrolling, the next, I’m sitting through an orientation that would flip my summer upside down."
It all started with a casual family dinner when my uncle, always full of surprises, casually dropped, “Hey, I know this amazing program you’d love!” Cue my curiosity skyrocketing. Little did I know, this one conversation would set me off on an adventure through the world of tech I never expected. When he mentioned Machine Learning and AI, I was hooked. It wasn’t just a program—it felt like unlocking a new level in a game I didn’t even know I was playing. And here I am, knee-deep in data, juggling code like it’s a circus act. Best. Tip. Ever.
Then there was Rohin, who had no idea his mom was about to change his plans entirely. “She told me about this cool internship that combined software development and data science. I looked into it, and the more I read, the more I knew this was something I had to do.”
And finally, Devin was just vibing with his extra summer classes when his roommate Siddharth planted the idea in his head. “AI/ML projects with industry mentors? I couldn’t pass that up.” And so, like dominoes, we all fell into this adventure.
Chapter 2: The (Not So) Smooth Beginnings
Alright, let's talk about expectations for a minute. You think you’re prepared for an internship, and then reality smacks you in the face like a misplaced semicolon in your code.
Diya and Devin here—I walked in expecting to do basic intern stuff: squashing bugs, writing code, and maybe fixing up some documentation. You know, the safe, background stuff while the “adults” handled the real tech. Instead, I found myself smack in the middle of complex research about which data sets (we scoured KRAGEN, MEDRAG etc) would be the most detailed, and bug free on GitHub. “Am I seriously doing this?” became my daily mantra.
Kunj felt the same kind of overwhelm but with a twist. “I was staring at these problem statements like they were written in an alien language. But once Niraj, our mentor, stepped in, things started to click. His chill approach made the impossible feel possible.”
For Siddharth, though, it was all about hype. “I had heard *so much* about machine learning like it was the crown jewel of tech. And suddenly, I was about to get my hands dirty with it. I couldn’t wait.”
And then there’s Rohin, who was just excited to learn anything and everything. “I didn’t care which challenge we took on. I knew, no matter what, I was going to come out the other side with way more knowledge than I went in with.”
Chapter 3: Being lost in the forest (only to hunt the perfect research idea)
The first time the four of us sat down to discuss what we wanted to accomplish, I distinctly remember feeling completely lost. None of us even knew what RAGS were, beyond what ChatGPT had explained to us. When Niraj mentioned he had experience in the medical field, we instantly knew that had to be the focus of our project. He answered our questions like he'd completed a PhD on the topic, while we were all still grappling with the basics. Every time I had a question, I couldn’t help but wonder, *Is this even valid?*—that's how unfamiliar the territory was for us. But then we embraced the classic advice: *take it one step at a time.*
That meant diving deep into research, spending countless hours on GitHub forking repositories, and eventually piecing together our work to create something far more complex than we ever imagined. During the execution phase, we divided tasks into pairs, which ended up being incredibly effective. It felt like a real-world "separation of concerns" approach—each pair focused solely on their portion of the project without interfering with the other, until the moment our code had to come together. That final integration was both literal and symbolic, as our individual contributions merged into something cohesive and meaningful.
Chapter 4: Cracking the Code (And My Sanity)
You know what no one tells you about internships? It’s one giant problem-solving puzzle with a lot of coffee and not enough sleep. The satisfaction when you finally figure out a bug after hours of trial and error? Yeah, that’s the good stuff.
For Diya (aka me), debugging became both my nemesis and my best friend. There was this one time I spent HOURS hunting down an issue, only to realize it was a misplaced CURLY BRACE. A “{}”. But the sense of accomplishment when I finally nailed it? Nothing beats that.
Kunj had a different kind of thrill—working on a mix of healthcare and the tech was something he hadn’t expected. “I loved learning how a Retrieval-Augmented Generation (RAG) system works with a knowledge graph. It was a game-changer for me, blending tech and medicine in a way that I never thought possible.”
Meanwhile, Devin was vibing on the team's energy. “It wasn’t just about writing code—it was about seeing how everything fit together, from brainstorming sessions to final product builds. Every time I pulled something off that I never thought I could do, it felt like leveling up in real life.”
Rohin found the most joy in teamwork. “For me, it wasn’t one specific part of the project; it was the whole process of collaborating with my team, learning from each other, and building something together that felt bigger than the sum of its parts.”
Chapter 5: The Time Management Black Hole
Let’s get real for a second: time management during this internship was a beast. I thought I knew how to juggle tasks—spoiler alert, I didn’t. Between coding, debugging, and keeping up with deadlines, some days it felt like everything was collapsing. Diya, once again, trying to balance multiple plates only to have a bug crash the whole stack. In the end, when we met twice a week for meetings, Niraj would sit with us for hours, patiently helping without ever making us feel incompetent (which was honestly a skill), only to fix an issue that took just five seconds to resolve.
Kunj took the struggle to a café, where late-night work sessions kept him company (along with copious amounts of coffee and the occasional slice of cheesecake). “It was tough, but watching the project slowly take shape, piece by piece, kept me going,” he says.
I still remember the first time we ran the command to launch the front end of our project—it was *so* exciting. Watching our chatbot actually respond to questions felt incredible, like taking that first sip of water after an unbearably long day without it. It was the perfect blend of relief and accomplishment, a moment that made all the late nights and endless debugging completely worth it.
Siddharth and Devin? Debugging was their personal villain. “It felt like every day I was fighting the same battle—chasing bugs and learning new things at lightning speed. It was frustrating, but honestly, that’s what made it exciting.”
And Rohin struggled with the learning curve more than anything else. “Reading through technical papers and trying to understand complex theories was no joke. But by the end, I felt like I had leveled up in a way that a classroom could never give me.”
Chapter 6: From Debugging to Dreaming Big
The learning outcomes from this internship were *no joke*. Diya here—I worked on integrating Chainlit with OpenAPI for a healthcare chatbot. Sounds fancy, right? Well, it was—handling real-time data and creating dynamic visualizations pushed my boundaries in ways I hadn’t expected. I also got to dive into using Git, Docker, LiteralAPI AI, Python, and NebulaGraph, which took things to a whole new level. One of the coolest aspects was color-coding the graphs to make insights more intuitive and using triples to extract and organize data efficiently. Kragen was a lifesaver for so much of this work!
During Kunj's internship with Garje Marathi on BioGraphRAG, he dove deep into enriching biomedical knowledge graphs and using metadata to enhance answers. He wasn’t just scraping data; Kunj was integrating gene IDs, protein structures, and drug information from databases like UniProt, AlphaFold, and RXNav, almost like building a treasure map out of data. He also conducted a node degree analysis to identify bottlenecks in retrieval, generation, and enrichment times, optimizing the system’s response time and efficiency. Throughout the project, Kunj utilized technologies such as RAG, Python, Docker, NebulaGraph, and LlamaIndex.
Siddharth became a multiprocessing wizard, mastering how to distribute workloads across multiple cores, which made tasks like querying large datasets and updating graphs in Nebula Graph far more efficient. "The whole concept of Graph RAG systems blew my mind. I was able to speed up processes from 35 hours to 3 hours by splitting the upsert of triplets into parallel tasks, reducing load times significantly. Watching how the system could handle complex queries and dependencies in real-time was like seeing a new world open up. It gave me a deep appreciation for the power of combining graph databases with natural language processing."
Rohin got his hands dirty with every aspect of the project, from setting up the knowledge graph to creating subgraphs for better data visualization. “It gave me a real appreciation for how all the moving parts come together to make something functional and impressive.” When setting up the knowledge graph, he had to upsert over 3 million biomedical entities, allowing for various generated responses to be more accurate and overall factually correct. Apart from the knowledge graph, he too worked on the code for generating the entities/triples for the user subgraph. By generating these triples it allowed for both Diya and Devin to more easily create the subgraphs for the users.
Chapter 7: The Non-Tech Wins
Now, let’s talk about the real lessons—the ones beyond lines of code. Diya came to understand how crucial communication skills are. Niraj always reminded us, “post your problems on Discord.” It became like our team mantra at one point, because we were all too scared of sounding incompetent to actually do it. But eventually, I realized there’s no such thing as a dumb question, and everyone has to start learning somewhere. My biggest takeaway was the importance of asking questions, and I’ve been applying that in my classes as well.
Kunj learned the value of mentorship. “Niraj’s calm, patient approach made it okay to make mistakes, to experiment. Having a mentor like that changed the whole experience for me.”
For Siddharth, teamwork and project management were the big takeaways. “Learning to sync up with the team, handle deadlines, and stay flexible was huge for me.”
Devin valued the collaboration more than anything. “Working as a cohesive unit instead of five individuals on separate projects made everything so much smoother—and more fun.”
For Rohin, collaboration was one of the most important aspects of the project. By sharing ideas and working in teams he was better able to understand the layout and overall mission for the project. “Collaborating during this internship was awesome, it made the project that much more fun while also improving it.”
Chapter 8: What’s Next?
What’s next, you ask? For Diya, it’s about merging creativity with technology to create meaningful, impactful projects. Devin, likewise, he's committed to making a difference through his work, regardless of how small the user base might be. As they put it, “Building something that improves someone’s life? That’s the goal.”
Kunj is passionate about working on cutting-edge technology, from AI explainability to groundbreaking robotics and neurotech. He seeks opportunities where the next major innovations are being developed.
Siddharth is looking for more internships—ideally, in cutting-edge AI projects. “I want to push myself even further and see where it takes me.”
And Rohin? “I’m keeping an open mind and diving into every niche of data science I can find—AI, NLP, computer vision, all of it.”
Chapter 9: To Future Interns
Here’s some advice for future interns (you’re gonna need it!):
Diya: Don’t hesitate to ask questions; COMMUNICATE; your mentor & teammates are your biggest resource.
Kunj: Embrace the journey—not just the destination—mistakes are lessons.
Siddharth: Debugging is going to take a significant amount of your time and patience, learn to enjoy debugging
Rohin: Collaboration is key; working together makes everything smoother—and more fun!
Devin: You can be whoever you feel content being.
Chapter 10: Words from the Mentor
I am Niraj Kumar Singh, the mentor to this group “code crafters”! I’ve spent the last few years working as an ML Engineer in the Bay Area, navigating the fast-paced world of startups. My focus has primarily been on ML engineering, particularly in LLMs and RAG systems, with a growing curiosity about adapting these technologies for healthcare. When my good friend Aakash Patil introduced me to Garje Marathi and invited me to mentor in their Summer of Code program, organized by MAAI, I was immediately intrigued. It felt like an opportunity not only to guide a team but also to tackle some of the latest challenges in LLM architecture through a fresh perspective.
Meeting this team for the first time, I could feel their excitement. They had the passion and curiosity that I believe are critical in tech, but more importantly, they had a hunger to learn. Most of them were stepping into real-world problem-solving for the first time, and I knew that while they had the coursework covered, the journey ahead was going to challenge their practical skills. But the team embraced the challenge head-on, often feeling like they were building a parachute after jumping out of the plane.
From literature reviews to forking open-source repositories on GitHub, the team learned to navigate uncharted territories—polishing their programming and debugging skills, working on cutting-edge tools like Claude and ChatGPT, and tackling Graph RAG systems with python libraries like Llama-Index and graph databases. This wasn't just about mastering technical concepts—it was about developing a resilience to tackle the unknown.
Watching them collaborate was perhaps the most rewarding part of the journey. I remember how they divided their tasks, dealing with everything from merge conflicts to handling both front-end and back-end systems. These were crash courses that will undoubtedly serve them well as they pursue internships and jobs. Their adaptability, commitment (pushing PRs at midnight!), and technical presentations impressed everyone—laying a solid foundation for the careers they’ll soon embark upon.
I encouraged them to not only focus on the technical skills but also to refine the soft skills that are just as critical in the workplace. Time management, communication, and working seamlessly in a team will all play vital roles as they continue their journey. Seeing how they took on these challenges without hesitation, even diving into new open-source projects, gave me confidence in their ability to succeed in the fast-paced world of tech.
Looking ahead, I’m excited to see how this team applies what they’ve learned. From RAG systems to healthcare tech, the doors are open for them to contribute meaningfully to any project they take on. I’m proud to have been part of their journey and can’t wait to see what the future holds for them.
A big thank you to the mentors and organizers at MAAI for their support in making this program a success. We’ve learned a lot in this first iteration and look forward to making it even better in the future. Stay tuned to Garje Marathi’s website for details on the next cohort!