UHG
Search
Close this search box.

How SuperKalam Uses OpenAI GPTs to Fuel UPSC Aspirants 

SuperKalam aims to go beyond UPSC and scale to cater to students attempting other competitive exams, including JEE and NEET, in the coming months. 

Share

Illustration by Nikhil Kumar

As the UPSC Civil Services Exam (CSE) date of June 16th draws closer, over 11 lakh aspirants are diligently preparing to tackle one of the world’s most challenging examinations. SuperKalam, an AI platform is empowering students with personalised learning experiences, paving the way for their success in cracking the UPSC CSE.

Vimal Singh Rathore along with Aseem Gupta, both of whom previously founded qoohoo, started SuperKalam in July last year. 

Vimal who cleared his UPSC in Central Armed Police Forces Exam (CAPF) exam but went on to work at Unacademy and then start his own company, Coursavy, which was acquired by Unacademy.

The duo recognised the need for a more adaptable and student-centric approach to UPSC preparation. With his extensive background in education and entrepreneurship, Rathore set out to create a platform that could bridge the gap between traditional coaching methods and the unique requirements of each aspirant.

In an exclusive interview with AIM, Rathore shared his insights on the platform’s genesis and its mission to revolutionise UPSC preparation. 

“The inspiration is pretty simple,” he explained. “There are three examinations in India which kind of change the trajectory of any student or any family’s future or life, which are UPSC, that is for the civil services, NEET to become a doctor, and JEE to become an engineer. These examinations cumulatively are given by 5 million aspirants every year.”

Rathore further emphasised the need for a more personalised approach to learning for these exams. He stated, “The entire responsibility of understanding, asking questions, whether they are grasping or not, is completely on the student. The platform somehow, because of this kind of process, was not accountable. They were unable to do a lot about how students can improve their learning outcomes.”

Leveraging OpenAI’s GPTs 

SuperKalam leverages multiple models, including Llama, OpenAI’s GPT-4, and GPT-3.5, to deliver personalised learning experiences. For generating MCQs, Super Kalam relies on OpenAI GPT-4, while reasoning-based problems are tackled using OpenAI GPT-4 Turbo.

Super Kalam assists students in creating personalised timetables, sending daily targets, and tracking progress throughout their learning journey. It helps resolve doubts, aids in mastering concepts, and can evaluate a UPSC student’s handwritten answer in less than 60 seconds—a process that typically takes 2-4 weeks in the current education market. 

The platform also conducts mock tests, identifies students’ strengths and weaknesses, and provides progress reports that foster self-awareness, unlocking their full potential.

Rathore revealed, “There are at least 14 Agents that we are using to understand what you are asking Super Kalam.” The platform also uses fine-tuning and prompt engineering techniques to enhance the accuracy and relevance of the generated content.

SuperKalam has invested in NVIDIA GPUs and cloud infrastructure to ensure scalability and cost-efficiency, although they declined to mention how many. 

“We are also leveraging NVIDIA GPUs for certain models to handle the performance” Rathore mentioned. As the platform’s user base grows, the team is exploring partnerships with local cloud service providers to optimise costs and performance.

Aspiring to become SuperKalam

Last year the founders were a part of the Y Combinator W23 batch and the company is backed by Gustaf Alstromer, a partner at Y Combinator. Rathore brings his experience as a founding team member and growth leader at Unacademy, while Gupta, the CTO, previously worked in the early engineering team at Razorpay.

Since its inception, Super Kalam has witnessed remarkable growth. The platform currently boasts a user base of over 46,000 students, with an impressive week 12 retention rate of 78%. 

Rathore proudly shared, “We launched this latest version of the product on 15th of March. So 355 students who attempted any number of questions that day, the total number of questions that you are seeing, is different for every student.”

The impact of Super Kalam’s AI-driven approach is evident in the success stories of its users. Rathore highlighted the journey of Navya, a UPSC aspirant who struggled with consistency and self-doubt. 

“Navya’s accuracy rate was 56% at 45 questions. And here, Navya is number one ranker, with 234 questions and 84% accuracy,” he beamed.

SuperKalam has set its sights beyond UPSC and aims to scale to cater to students attempting other competitive exams. “At the same time, our mission is to make quality education accessible and affordable,” Vimal concluded. 

📣 Want to advertise in AIM? Book here

Picture of K L Krithika

K L Krithika

K L Krithika is a tech journalist at AIM. Apart from writing tech news, she enjoys reading sci-fi and pondering the impossible technologies, trying not to confuse it with reality.
Related Posts
19th - 23rd Aug 2024
Generative AI Crash Course for Non-Techies
Upcoming Large format Conference
Sep 25-27, 2024 | 📍 Bangalore, India
Download the easiest way to
stay informed

Subscribe to The Belamy: Our Weekly Newsletter

Biggest AI stories, delivered to your inbox every week.

Flagship Events

Rising 2024 | DE&I in Tech Summit
April 4 and 5, 2024 | 📍 Hilton Convention Center, Manyata Tech Park, Bangalore
Data Engineering Summit 2024
May 30 and 31, 2024 | 📍 Bangalore, India
MachineCon USA 2024
26 July 2024 | 583 Park Avenue, New York
MachineCon GCC Summit 2024
June 28 2024 | 📍Bangalore, India
Cypher USA 2024
Nov 21-22 2024 | 📍Santa Clara Convention Center, California, USA
Cypher India 2024
September 25-27, 2024 | 📍Bangalore, India
discord-icon
AI Forum for India
Our Discord Community for AI Ecosystem, In collaboration with NVIDIA.