It’s early morning at the Rajarambapu Institute of Technology campus in Sangli, Maharashtra. The corridors are still waking up, but inside a computer lab, three young engineers are already at work — fine-tuning code, testing responses, and training an AI system to do something deeply human: support.
For 21-year-olds Vaishnavi Rajkumar Patil, Abhishek Shivprasad Patil, and Ganesh Ramchandra Mahadik, collectively known as Team Neurostars, this is a personal mission.
Their AI-powered chatbot, designed to help parents of children with autism navigate everyday challenges, recently won GRASP 2026 — a national AI hackathon conducted in partnership with ASME India.
Competing against over 5,000 students, the trio secured first place in the “AI for Social Good” category, earning Rs 1 lakh, mentorship opportunities, internships, and a chance to collaborate with a European university.
Their work signals something worth paying attention to: young engineers in India choosing to build for people first.
Three villages, one lab
All three students come from small towns and agricultural families in Maharashtra, where life is shaped as much by the land as by aspiration.
Vaishnavi, who grew up in Karve village, recalls how her interest in technology began during the pandemic while observing her cousin sister’s work-from-home routine.“I was just curious about what she was doing and what these codes were, and gradually I started learning from her and decided to pursue computer science,” she tells The Better India.
From farming families in rural Maharashtra to a national AI stage, Team Neurostars is redefining what student innovation can look like.
She lives with her parents, younger brother, and grandparents, all of whom share the weight of farming and daily life. “My father works as a quality testing engineer, but along with that, he also looks after our farm; my grandparents are more into agriculture,” she adds.
Abhishek, from Kavalapur, traces his journey back to a childhood fascination. “I got my first computer when I was in Class 7, so I decided that when I grow up, I would pursue computer science engineering,” he shares.
At home, responsibilities are split between business and farming. “My father owns a shop, and along with that, he looks after our farm, and my mother is a homemaker.”
Ganesh, from Nevari, followed a similar path, growing up in an agricultural household before stepping into engineering. He now works as an intern while continuing to build on the project.
What unites them is not just their academic path but a grounded understanding of everyday challenges, where access to resources is limited, and solutions need to be practical.
A classmate’s story changed everything
The trio’s journey into autism awareness began with a friend’s story.
Through the IUCEE (Indo-Universal Collaboration for Engineering Education) student club at their college, the three were encouraged to engage with real-world problems. And one day, a classmate opened up about his sister’s autism and the toll it was taking on his family.
“Until entering engineering, we had no understanding of autism,” Vaishnavi admits. “In the village, people are not very aware, and they are not ready to accept it.”
She continues, “Through a friend, we learned about autism and how his sister was affected, and how the family was struggling to deal with it.”
Hearing their friend’s experience up close changed something. What had been a distant, unfamiliar condition suddenly had a face, a family, and a daily reality they couldn’t look away from.
Through their initiative and NGO, the team has been taking autism awareness into schools and communities where conversations are often missing.
In 2022, the three launched a youth-led initiative called ‘United for Autism’ with their peers, taking awareness campaigns into schools and local communities. Over time, this evolved into a registered NGO, ‘Dream Udaan Foundation’, which is the formal extension of their early initiative. Today, Vaishnavi serves as CEO, Abhishek as CTO, and Ganesh as a core member, continuing their work through both community outreach and technology.
But the deeper they went, the more they saw what was missing.
The problem no one was solving
Autism care in India often comes with structural challenges: limited therapists, high costs, and infrequent sessions. And the numbers are only growing. “When we did our research, we got to know that after the COVID-19 period, the number of autism cases increased by 306%,” Abhishek says.
Yet even as the need surged, access to care remained inconsistent. “Therapies were happening weekly or monthly, but parents face problems daily,” Vaishnavi adds. These daily challenges often include managing behavioural outbursts, navigating communication gaps, and supporting a child’s emotional regulation — all without continuous professional guidance.
That gap, where parents are left without guidance between sessions, became the foundation of their idea. “So we thought, as technical students, we could create something to do our bit to make autism care easily accessible and affordable,” Abhishek says.
Their answer: an AI-powered chatbot designed as a 24/7 therapeutic companion.
‘Mentorship took us from idea to prototype’
While the idea came from community work, turning it into a research-backed solution required guidance and persistence. “The biggest challenge was technological. We had the idea, but we didn’t know which tools to use,” Abhishek recalls.
Their head of department played a key role in shaping their approach. He said that it was okay to give answers, but ChatGPT was also giving answers. So what could be the uniqueness in our solution?” Vaishnavi shares. “So he told us to research more.”
That push sent them down a rabbit hole of research. They began exploring technologies like Retrieval-Augmented Generation (RAG) and agentic AI, looking for ways to make their chatbot smarter and more reliable.
Designed as a 24/7 companion, the chatbot helps parents navigate daily challenges between therapy sessions with real-time guidance.
Along the way, their understanding of autism was also strengthened through collaborations with organisations like Dream Udaan Foundation and Jai Vakeel Foundation, while therapist Dr Ashwini Patil guided them during testing and validation.
A Google-supported hackathon gave them another boost: a month-long mentorship and access to Google Cloud tools that helped sharpen their technical approach.
Piece by piece, the idea took shape into a working prototype.
Designing a companion for everyday challenges
The chatbot started as a straightforward idea. “When we began, we thought of just providing a chatbot,” Abhishek says. “But that was at the academic level.”
As they spent more time with families, they realised that every child’s needs are different, and the solution had to reflect that. So they introduced an AQ test within the application, a standardised clinical screening tool that they adapted to generate more personalised responses and insights.
“When parents answer those questions, it generates a score. If the score is less than 20, it means low, between 20 and 30 is medium, and above 30 is high,” Abhishek explains. “Based on that, we can tone the answers.”
Then came a second shift. Therapists they were working with highlighted the everyday struggles parents face between sessions, from handling sudden meltdowns to managing unpredictable sleep patterns. To address this, the team built a daily behavioural tracker.
“We identified five key behaviours in children: meltdowns, sleep patterns, communication, mood, and sensory triggers, and built a daily tracker around them so parents can consistently monitor and better understand their child’s needs,” Vaishnavi says.
Parents log simple daily inputs, which the system converts into weekly insights. “Then, it will take all these five things, create a dashboard for the particular child, and give insights based on their week,” she adds.
These reports help parents spot patterns and can be shared with therapists for more effective sessions. The chatbot also provides real-time support alongside the tracker. “If a child is facing any issue and the parent does not understand what to do, they can ask their query, and it will respond in a second,” Abhishek says.
Both features sit within a single web-based application called ‘DreamConnect’, designed to bring continuous, accessible support into one place.
The team is clear about one thing, though. “It is not replacing the therapist; it is just working as a companion to the therapist between therapy sessions,” adds Ganesh.
The technology behind the chatbot
DreamConnect is built across three layers: frontend, backend, and data. The frontend uses React.js, while the backend runs on Python to handle logic and AI communication.
At its core is a system called Retrieval-Augmented Generation, or RAG. Instead of generating answers from scratch, it pulls relevant information from a curated database first and then responds.
“RAG provides more specific answers,” Vaishnavi explains. “In LLM, there is a risk of hallucination, but RAG gives accurate answers based on data.”
In practice, this is what happens: a parent types a query, which gets converted into a numerical format using Sentence Transformers (MiniLM-V2), matched against stored information through a tool called FAISS, and then processed alongside the child’s AQ score using Gemini 2.5 Flash to generate a tailored response.
With features like AQ-based responses and behaviour tracking, the tool translates everyday observations into meaningful insights for families.
What makes this reliable is where the data comes from. “We have trained it on frequently asked questions collected through community work, inputs from Dream Udaan Foundation, and resources contributed under the guidance of therapist Dr Ashwini Patil,” Vaishnavi says. FAQs gathered through Nirmal Hospital in Miraj have also been used to strengthen the responses.
Parents log daily observations on the app, which are stored using Firebase and turned into visual weekly reports using Chart.js, giving families and therapists a clear picture of how a child is doing over time.
Winning GRASP 2026: A turning point
Their breakthrough came at GRASP 2026, a national AI hackathon organised by KRUU, a student innovation platform, in partnership with ASME (American Society of Mechanical Engineers) India. The hackathon was focused on solving real human problems through reasoning-led innovation.
Unlike traditional hackathons, GRASP emphasised research, empathy, and real-world application.
Competing against over 5,000 participants, Team Neurostars stood out for both their technical depth and human-centred approach. Their win brought Rs 1 lakh in prize money, national recognition, internship opportunities, a chance to collaborate with a European university, and mentorship to scale their solution.
For the team, it validated not just their idea, but their belief that technology can be deeply human.
A shared dream, still growing
The chatbot is currently a prototype, yet the team is already planning its next phase.
“With the help of KRUU and the India Autism Center, we are thinking about deploying this as scalable software,” Vaishnavi says. Multilingual support is also in the works, so the tool can reach families beyond English-speaking communities. This collaboration began during their post-hackathon mentorship phase, where conversations around real-world deployment and scaling first took shape.
Winning GRASP 2026 validated their belief that technology, when built with empathy, can bridge real gaps in care.
The three are building beyond the chatbot, too. As co-founders of Dream Udaan Foundation, with Vaishnavi as CEO, Abhishek as CTO, and Ganesh as Technical Lead, they want to use the NGO as a launchpad for AI-driven social impact.
“We want to improve this and work for this NGO, using AI for social good and sustainability,” Vaishnavi says.
Abhishek sees a bigger horizon. “We want to grow this small project into a big startup, which will be recognised all over the world.”
And Ganesh keeps coming back to the question that started it all. “Everyone is thinking about the future of AI, but I want to work on the future of humans using ethical AI.”
Back in that campus lab in Sangli, the code continues to evolve. But the three of them already know what they are building towards: a world where a parent sitting in a small town, unsure of what to do next for their child, can open a phone and find help waiting.
That is what brought them here. And that is what keeps them going.
All images courtesy Team Neurostars
