As a Second-year B.Tech student majoring in Artificial Intelligence and Machine Learning, I was thrilled to land my first summer internship at Persist Ventures, a cutting-edge company developing AI-powered productivity tools, solutions and projects in many different domains. Little did I know that these three months would be a whirlwind of learning, challenges, and growth that would shape my understanding of AI applications in the real world.

The Journey Begins

On April 9, 2024 I started working at Workplete (https://workplete.com/) Project of Persist Ventures and a mix of excitement and nervousness coursing through me. The team's warm welcome immediately put me at ease, and I was quickly introduced to the one of the two main projects I'd be working on: an Automatic Form Filler.

Project 1: Automatic Form Filler (AFF)

My first task was to develop a Chrome extension that could automatically fill out online forms across multiple websites. Sounds simple, right? Well, not quite.

The challenge was to make it intelligent enough to understand different form structures so that it can work around any sort of website with forms and accurately map user-provided information to the correct fields. The learning curve was steep. I had to quickly get up to speed with Chrome extension development, dive deep into the world of GPT-4 for intelligent field mapping, and learn how to optimize API calls to keep costs down. One of the biggest hurdles was handling the context length for GPT-4. The solution? Implementing FAISS Vector DB to reduce token length. It was my first time working with vector databases, and it opened up a whole new world of possibilities.

Seeing the extension's accuracy improve from 40% to 76% for the given sheet of different form links over the development period was incredibly rewarding. Even more gratifying was the positive feedback from beta testers, who reported saving a lot of time filling repetitive forms. It was my first taste of creating something that directly impacted users' lives.

Project 2: Multi-Action Agent (MAA)

Just as I was getting comfortable with the AFF, I was thrown into the deep end with the Multi-Action Agent project.

This was next-level stuff — an AI-powered browser assistant that could perform complex, multi-step tasks based on natural language commands. The MAA pushed my skills to the limit. I had to grapple with natural language understanding, autonomous web navigation, and visual understanding of web pages using GPT-4's vision capabilities. The Chrome Debugger API became my new best friend (and occasional nemesis) as I worked on giving the agent more control over the browser.

One of the most exciting moments was when we successfully tested the MAA on tasks like booking a cab on Uber or sending a WhatsApp message —all initiated with a simple text command. It felt like we were building the future of web interaction.

Beyond Technical Skills

While the technical skills I gained were invaluable, the soft skills I developed were equally important. I learned how to communicate complex technical concepts with the team. Team meetings and code reviews improved my ability to give and receive constructive feedback. One of the most memorable experiences was presenting our projects to potential investors. It was nerve-wracking but exhilarating, and it gave me a glimpse into the business side of tech.

Reflecting on the Journey

As I wrapped up my internship on July 1, 2024, I couldn't believe how much I had learned in just three months. From Chrome extension development to AI integration, from database management to UX design considerations, the experience was comprehensive and eye-opening.

But beyond the technical skills, this internship taught me about the real-world application of AI. It showed me how theoretical concepts I'd learned in university translated into products that could genuinely improve people's lives. It ignited a passion for creating AI-powered tools that are not just technically impressive, but also user-friendly and practical.

To all students considering internships: take the leap. The learning, the challenges, the victories — they're all part of an incredible journey that will shape your future in tech.

As I head back to complete my degree, I carry with me not just new skills and knowledge, but a renewed excitement for the future of AI and my role in shaping it. The journey has just begun, and I can't wait to see where it leads next.

This internship was more than just a summer job. It was a glimpse into the future — both the industry's and my own. And if this is just the beginning, I can't wait to see what comes next.