From libraries and laboratories, to deep friendships, social activism and insightful research, IIIT-Hyderabad alumna Deepti Ghadiyaram shares the trajectory and life choices that flavour her research in building safe, interpretable, and robust multi-modal AI systems.

In a world, where humanity consumes endless, sometimes mindless content online, Deepti Ghadiyaram’s computer vision research focuses on healthier and robust AI systems. 

The assistant professor of Computer Science at Boston University and affiliate faculty in ECE, Computing & Data Sciences was a former senior Research Scientist at Meta AI and Runway. She secured her Ph.D from the University of Texas at Austin in 2017. 

Rooted in a Culture of Learning
Deepti grew up in a bustling and intellectually nurturing joint family in Hyderabad. It was here that her maternal grandfather first encouraged her empirical curiosity. As the younger sibling, she spent her early years diving into her older brother’s collection of comics and science fiction novels—fueling the imagination of a young mind.

“My mom, a chemistry professor at Vasavi College, heard about the relatively new IIIT Hyderabad from her colleagues,” Deepti recalls. “They described it as a premier institute with eminent faculty and a research-focused curriculum. It was a win-win. I could experience campus life while staying close to my family.”

Finding Her Calling in IIIT Hyderabad 
Despite having no prior exposure to programming, Deepti credits Prof. C. V. Jawahar’s computer programming course to be a pivotal challenge to overcome. It led to countless hours in the campus computer labs. “I had a full circle moment when I conveyed how much work it must have taken on Prof. C. V. Jawahar’s part in designing that course and brainstorming on effective teaching strategies with him during my last visit to IIIT-H”. Her interest in visual media was sparked by digital image processing with Prof. Jayanthi Sivaswamy and computer graphics with Prof. P J Narayanan. She also fondly remembers the unmatched style and timely guidance of Prof. Govindarajulu.

“The seed for research was planted when I realized I could take an image, program an algorithm, and transform it into something else,” she says. The reward of research became even clearer in her third year when she and her friend had their first publication accepted at ICVGIP. “IIIT-H’s research-focused curriculum is unique and nurturing,” she notes. “I recently gave a talk at CVIT and was blown away by the students. Their brilliance and engagement filled my heart with joy.”

Beyond academia, Deepti found a lifelong support system. “I found a small circle of like-minded friends who are still my closest confidantes today. I have such great memories of endless walks around the green campus, grabbing samosas at the canteen, and staying up late watching movies.” She would connect with batchmates at alumni meet ups, when IIIT-H faculty visited the Bay Area. 

Samvedana: connecting communities with compassion
Towards the end of her fourth year, Deepti got deeply involved in social entrepreneurship on campus. She founded Samvedana, an NGO that connected less-funded organizations to financial aid from corporates, friends and seniors in high paying jobs. “We financed students and NGOs from lower income groups. It was heartwarming to see how so many of my friends came together, and selflessly volunteered their time and talent, designed, built, and maintained a website for years, went door to door in hot summer, and helped document not very well known old age homes and orphanages”. 

“It was very hard to run and scale this organisation in a self-sustained manner. After about 18 months, my father pointed out that I should probably not give up on my passion for research at a relatively young age”.

Ph.D as a milestone in confidence
In 2011, Deepti moved to the University of Texas at Austin for her Master’s in Computer Science. Though she initially doubted her ability to pursue a Ph.D, her dedication and contributions to her research group led to a doctoral offer from her advisor.

“I was initially mortified about going at it alone, since my parents were not yet convinced,” she admits. “They came around eventually but those early days were emotionally rough. I was lucky to have a strong support system of like-minded friends and an advisor who compensated with liberal doses of positivity. We were building algorithms to detect if real-world images and videos were perceptually of good quality or not.” 

Crossing that Ph.D milestone definitely boosted my self-confidence”, notes Deepti who received the UT-Austin’s Graduate Recruitment Fellowship (2013-2016), offered to the top 10% of students by the Department of Computer Science. “When I was on the other side, I saw the quality of applicants and it blew my mind that I was one of the recipients”, laughs the researcher,  with over 50 publications who has co-authored two books.

From Industry to Academia
Deepti’s career trajectory took her to Facebook AI, where she built algorithms to detect objectionable content on Facebook and Instagram, and later to Runway, a generative video startup. She credits her early exposure to fundamentals and coding competitions at IIIT-H for her success in these high-stakes environments.

Today, as a faculty member at Boston University, Deepti “wears many hats”—from teaching and mentoring to sourcing funds and presenting research. “I’m proud of my ability to context-switch very quickly,” says the poster girl for the non-extrovert.

“At Boston University, I do a little bit of everything, from video generation to model evaluation”. Deepti explains, “When software is deployed in robotics or serious applications, we should fully understand the errors that models can make”. 

She received the Moorman-Simon Interdisciplinary Career Development Professorship award from Boston University (2025-2028) for two of her projects.  Her current work is deeply impactful.  The AI assistant tool for paediatric emergency care is an ongoing effort with a paediatric emergency doctor. The intention is to build trauma-care protocols that can be administered immediately by first-responders till the young patient is delivered to medicare professionals. The sign-language tool seeks to create more inclusive classrooms, with a digital assistant to translate what the teacher is saying in real time. 

What sparks joy
Despite her high-reaching career, Deepti views her arc as a collective effort powered by the love of her family and friends. She is married to fellow IIIT-H alum Rohit Gernapudi, whom she met during their days volunteering for Samvedana.

Today, their world is ruled by their three and half year-old daughter, Mahira. The family finds joy away from screens, preferring greenery and quiet water bodies. When asked about a recent pleasure, she paused and said, “oh, jumping into muddy puddles with my daughter at a neighbourhood lake.”

While her reading list now leans toward human psychology and history, her media tastes remain happily “stuck in the 90s” with re-runs of Seinfeld, Curb Your Enthusiasm, and Portlandia. “I used to attend many live concerts in my past life but today it is only Disney movie reruns for us,” she laughs. 

“What gives me the most joy professionally is seeing my students grow,” she reflects. “When a student tells me that something I taught was useful or inspired them to pursue research, or that my recommendation letter helped them get into their dream school—those are my unmatched pearls of happiness.”