It was a usual late evening—me at my desk in the Centre for Innovation at IIT Madras, slogging through my 4th semester final exams. Suddenly, an email pops up:
“Congratulations! You have been selected for admission to the 2025 NUS Young Fellowship Programme.“
I blinked. Then I read it again. And then I jumped. You see, I had been hoping to spend my summer doing something meaningful, and this? A fully funded, week-long program at the National University of Singapore, with 100 exceptionally talented peers from 22 countries, all coming together to discuss AI and its impact on the world? This was everything I could’ve hoped for—and more.
What followed was a deeply enriching journey into the future of AI: how it’s shaping everything from medicine to music, business to biology. We met with faculty from all across NUS—engineering, computing, psychology, science, even the arts—and saw firsthand how AI is revolutionizing their fields. From LLMs and generative models to psychological studies on AI’s societal effects, the week was a whirlwind of ideas, perspectives, and inspiration.
But the experience wasn’t just about lectures and learning—it was about people. I had the chance to connect with brilliant, passionate peers from over 22 different countries. From late-night dorm chats to lunch table debates, from strolling through Gardens by the Bay to getting lost in the food courts of Chinatown—we talked about research, life, the future, and everything in between. What struck me most was how diverse our backgrounds were, and yet how aligned we were in curiosity and ambition.
Hanging out with them was energizing. We joked about our home universities, helped each other brainstorm poster ideas, took turns translating things on signs, and shared music, snacks, and stories from back home.
Oh, and the best part? I also got to explore the beautiful, vibrant, meticulously clean country of Singapore along the way. Every evening after the day’s sessions, I was out walking—20,000 steps a day at least—checking off my bucket list, soaking in the skyline, and experiencing a new place with new people and fresh eyes.
Before the Flight: A Plan to Make the Most of It
By the time the last week of June arrived, I was boarding my flight with a notebook full of ideas and a very long personal itinerary. I mean, if you’re getting a fully funded trip to Singapore, why not make the most of it? The program ran from 9 to 5—leaving the rest of the day wide open. And I wasn’t about to let that go to waste.
Every day of my 10-day stay was meticulously planned to be equal parts productive, exciting, and joyful. I don’t think there was a single evening I returned to my dorm without clocking in 20,000 steps and legs begging for mercy.
The NUS Young Fellowship Programme
We kicked off with a warm welcome from Prof. Jessica Pan, a group photo with fellows from around the world, and a debrief by Prof. Chai Kah Hin on this year’s central theme:
How does generative AI affect PhD research?
What followed were five days of tightly packed, thought-provoking lectures, workshops, and discussions. Here are some of my favorites:
What Is Generative AI?
Prof. Amirhassan Monajemi opened the floor by walking us through a historical timeline—from Boltzmann machines to AlexNet, GANs, transformers, and today’s LLMs and agentic AI. He highlighted a critical shift: we’ve moved from “simulating intelligence” to “learning intelligence”.
While the technological progress is undeniable, we also dove into the often-overlooked challenges—the ethical, political, and interpretability concerns that come with this power. Deepfakes, algorithmic bias, job displacement, misinformation—these aren’t side effects, they’re central to how AI integrates into society.
One point that really stuck with me was skill degradation—how relying too heavily on AI might subtly erode our own abilities. For instance, there’s a recent MIT study showing how students using LLMs for writing may produce better essays in the short term, but risk losing writing depth over time. It’s a subtle shift, but an important one—and it made me pause and ask myself: how much of this blog am I really writing on my own?
Data-Centric AI for Language Models
This session by Prof. Bryan Low was an eye-opener. He shared a compelling argument:
Training LLMs is fundamentally a data-centric problem.
In a world where “garbage in, garbage out” holds truer than ever, why not focus on optimizing the data pipeline? His lab showed how using just 5% of the full dataset, carefully curated, can outperform training on 100%. This is what their work NICE explores—cutting redundancy and improving performance through smarter data selection.
We also explored:
- DUET, ActiveDPO, Top-m: methods for data filtering and optimization
- DETAIL, Freeshap, Waterfall, WASA: tools for data attribution and provenance
- Machine unlearning: a fascinating reverse problem—how to remove information from AI models when needed (e.g., for privacy or copyright)
Check out glow.ai for more on his lab’s work—genuinely fascinating stuff.
Assistive Technology and Thinking Outside the Box
Prof. Suranga Nanayakkara’s talk was a favorite of mine. It wasn’t about raw technical horsepower, but about designing with empathy. He showed how his lab reimagines assistive tech: tools that adapt to humans rather than the other way around.
Some amazing projects to check out:
- SPARSH: tactile interfaces
- EARPUT: turning the ear into an input surface
- FINGERREADER: wearable reading aid for the visually impaired
- AiSee: smart assistive vision
He encouraged us to think differently. His advice:
- Reframe the problem – don’t ask “how to make a better alarm clock” but “how to ensure people wake up on time.”
- Challenge assumptions – why rely only on vision when other senses might work better?
- Pay attention – most people tolerate poor solutions; innovators notice what others ignore.
- Make it happen – because talking without building is just noise.
Also: he introduced us to chindogu—“unuseless inventions” that are technically usable but wildly impractical. A fun way to rethink creativity, if you ask me.
Machine Learning for Predictive Biodesign
This one was new territory for me. Coming from a non-biology background, I found Prof. Matthew Chang’s lecture on AI in enzyme engineering and biomanufacturing fascinating.
We explored how machine learning is used in DBTL (Design-Build-Test-Learn) cycles for things like:
- Designing genetic constructs
- DNA editing
- High-throughput testing
- Pattern discovery
From protein GANs and ESM models to a case study on cannabinoid biosynthesis, this opened my eyes to a new application of AI in synthetic biology—designing cells, not just code.
AI and Robots in Our Human World: A Psychological Lens
Prof. Sam Yam gave a much-needed shift in perspective. While we tech folks are busy tuning models and optimizing performance, it’s easy to forget the human response to all this.
He asked: Can AI fail us?
He shared the story of a medical AI that was racially biased against Black patients—because it was trained to optimize healthcare costs, and cost patterns themselves reflected societal biases.
From robots as coworkers (and threats) to priests and therapists, we studied psychological studies showing AI’s mixed effects. It’s not all bad—or all good. It’s contextual. It’s complicated. And it made me realize the importance of taking a step back from the code to understand the human narratives surrounding AI.
How Does AI Affect PhD Research?
As part of the program challenge, we formed teams and prepared posters and 3MT (3-Minute Thesis) presentations. My team worked on the concept of an AI collaborator in research—somewhere between a human scientist and a fully automated system like Sakana.ai.
We proposed a pipeline that enables an AI to collaborate with a human researcher, offering ideas, verifying claims, and even contributing iteratively. We argued for AI as a thinking partner—not a tool, not a threat, but a co-researcher.
Other teams came up with equally fascinating ideas. One that particularly stood out to me was the idea of reframing AI as a mirror—not a threat, but a challenge that forces you to grow beyond what machines can replicate. Their message was sharp and thought-provoking: “If ChatGPT can do the research you’re pursuing, then maybe you’re not asking the right questions.”
It wasn’t just a critique—it was a call to aim higher. To push toward creativity, depth, and originality that transcend what’s easily automated.
Outside the Lecture Halls: NUS, Singapore, and Everything In Between
Of course, the academic side was just one part. We explored the NUS campus, toured its labs, and spent the evenings and weekends soaking in everything Singapore had to offer—Gardens by the Bay, Sentosa, Chinatown, Little India, the museums, the food, and that magical late-night skyline from Marina Bay Sands.
I’ll attach some pictures at the end as a mini gallery of memories.
Final Thoughts
It’s hard to compress the richness of this experience into a single blog post. The NUS Young Fellowship wasn’t just about AI—it was about people, perspectives, places, and possibilities.
It made me reflect more deeply on my research direction. It encouraged me to embrace cultural diversity and different ways of thinking. And it inspired me to seek out more global opportunities—programs, internships, and conferences abroad. Because when you’re working on something meaningful, with the right people, in the right place—it doesn’t feel like work. It feels like an adventure.
But beyond all the lectures, discussions, and lab visits—there was something deeper. I was living life fully. I was researching something I cared about, surrounded by brilliant people from around the world, in a completely new country. Evenings were spent wandering through Singapore’s colorful neighborhoods, catching the surreal light shows at Marina Bay, and laughing over late-night snacks with new friends.
This week taught me a life lesson I’ll carry forward: work hard, chase what excites you, and say yes to opportunities that take you across borders—intellectually and geographically.
Because the real joy? It’s in the blend—grinding through your academics, earning these chances, and then pausing to explore, to feel, to live.
And hey—when that journey is covered by a fellowship and sprinkled with a bit of magic? It’s kind of like a funded vacation with a little research on the side. 😉