Agha Ali Raza

Detailed Bio

Dr. Raza aims to enable information access and connectivity for under-connected, under-served, and offline populations (low-literate, low-income, tech naive, visually impaired, linguistically/socially marginalized, geographically remote communities, and oral cultures -- link to TEDx talk). To this end, he has developed and deployed novel systems and services. These services include a viral speech-based information dissemination platform accessible over basic phones called Polly, a maternal health and vaccination hotline for expectant fathers called Super Abbu, a voice-based social network called Baang, a voice-based survey platform called Sawaal, and a voice-based crowd-sourcing platform called Karamad. These inclusive services allow users to interact over phone calls from basic (non-smart) phones. These services have reached over 300,000 low-literate and non-tech-savvy users in Pakistan, India, and West Africa. They have provided instrumental information on maternal health and childcare, COVID-19, Ebola (in West Africa), routine immunization, job opportunities, and weather. His work in ML and Language Technologies focuses on tailoring and democratizing these technologies for the developing world and resource-constrained settings. He developed the first medium vocabulary speech recognition system for Urdu in 2009. He also developed tools, including Urdu text-to-speech, spoken term detection, speech emotion identification, voice biometrics, and audio deepfake detection (1, 2, 3) and techniques, including supervised and unsupervised pruning(1, 2), to bring large datasets within the computational reach of moderate computing resources. He also released the first open-type, character-based Urdu Nastaliq web font in collaboration with Mr. Nasrullah Mehr, a famous calligrapher. His work on the nexus of ML and Social Good ensures safe and ethical technologies by proactively identifying and mitigating the social impact of potential harms like misinformation and deepfakes, especially among tech-naive and low digital literacy populations. He draws on methods from various disciplines, including mixed methods (qualitative and quantitative), user studies, surveys and focus groups, RCTs, and participant observations.

Dr. Raza has authored 44 research papers (30 post-Ph.D.) published in peer-reviewed conferences and reputable journals, including CHI - the top publication venue in HCI (7 papers), CACM (IF: 14, 2 papers), InterSpeech - the top conference in speech processing (4 papers), WWW, CSCW, and COLING. These include nine CORE-A*, eight CORE-A, and eight journal papers. His novel approach to mass-disseminate instrumental information to low-literate people in Pakistan using entertainment as a hook created a paradigm shift in information services for under-served populations and received the best paper award at CHI. His work on developing a unique method to disseminate maternal health information to expectant parents in developing countries received an Honorable mention at CHI. His students and RAs have co-authored several of these publications. He has been awarded highly competitive grants by the UNICEF Innovations Fund, GIZ, Facebook/Meta, the Gates Foundation, the National Institutes of Health (NIH), the National Academies of Sciences (NAKFI), Google Inc., and the Higher Education Commission of Pakistan. As a lead PI from Pakistan, he has received grants worth USD 1,574,500 from these prestigious organizations. He has been an Associate Chair for CHI since 2019, the short papers PC chair for ICTD 2017, the open session chair for ICTD 2022, and the Associate Editor for the ACM Journal on Computing and Sustainable Societies (JCSS).

Dr. Raza aims to democratize knowledge. To this end, he has made his ML, Speech Processing, and Computing Fundamentals courses public, which have been watched more than 90,000 times with notes of thanks from eager learners. He has recorded Physics and Mathematics lectures for eighth graders with the Punjab E-learn project. With Edkasa (an education technology startup) and TikTok, he co-developed educational videos aimed at helping students prepare for their exams. The campaign, dubbed #examready, has been a massive success, with TikTok users watching the videos over 750 million times and over 100,000 videos created with the #examready hashtag. He has also developed an online Machine Learning course with LUMSx, the distance learning wing of LUMS, to be released over the Summer of 2023.

Dr. Raza has strong research collaborations with CMU, the University of Washington, the University of Michigan, ICSI, the University of Cape Town, the Universities of California - Davis and Riverside, and Cornell. He regularly gets invited to deliver guest lectures and keynote addresses at prestigious venues, including Meta, Google Inc., YouTube, Microsoft Research India (MSRI), NYU - Abu Dhabi, CMU - Qatar, and the UNESCO Chair in Tech4Dev at EPFL. He has delivered guest lectures at Cornell University, Lorentz Center, University of Maryland, Tufts, University of California - Berkeley, and the University of Michigan. Recently he has been conducting workshops and talks to spread awareness about Large Language Models (LLMs), Artificial General Intelligence (AGI), and ChatGPT and the tremendous potential of such technology to transform our lives and society (link). He has also been delivering talks to sensitize the community to the need for responsible AI and the risks of technology, including misinformation and deep fakes. His projects have been featured in prestigious national and international media, including BBC, NPR, MIT Tech Review International, Express Tribune, Dawn News, Geo News, The Times of India, IEEE Signal Processing Society, Pittsburgh Post-Gazette, and the Voice of America.

Short Bio

Dr. Agha Ali Raza is an Assistant Professor of Computer Science at LUMS, a Mahbub ul Haq Research Center (MHRC) fellow, and a member of the consortium of experts that leads the MS in Artificial Intelligence program. He is the founding director of the Center for Speech and Language Technologies (CSaLT) and the Principle Investigator of the Crime Investigation and Prevention Lab (CIPL) under the National Centers of Big Data and Cloud Computing (NCBC) initiative of the government of Pakistan. He is a Fulbright Doctoral Scholar and received his Ph.D. in Computer Science with a specialization in Language Technologies from the School of Computer Science at Carnegie Mellon University, USA. His cumulative experience as part of the Computer Science faculty at FAST-NU, ITU, and LUMS spans thirteen years. He teaches Computer Science courses at the graduate and undergraduate levels, including Natural Language Processing, Speech Processing, Machine Learning, Data Structures & Algorithms, Assembly Language Programming, Digital Logic Design, and Computer Architecture. Dr. Raza works to help foster an accessible, inclusive, equitable, and safe digital world where everyone has the opportunity to thrive. His research interests span Speech and Natural Language Processing, Speech-based Human-Computer Interfaces, Machine Learning, and Information & Communication Technologies for Development (ICT4D). Dr. Raza aims to enable information access and connectivity for under-connected, under-served, and offline populations (low-literate, low-income, tech naive, visually impaired, linguistically/socially marginalized, geographically remote communities, and oral cultures). To this end, he has developed and deployed novel systems and services. These services include a viral speech-based information dissemination platform accessible over basic phones called Polly, a maternal health and vaccination hotline for expectant fathers called Super Abbu, a voice-based social network called Baang, a voice-based survey platform called Sawaal, and a voice-based crowd-sourcing platform called Karamad. These inclusive services allow users to interact over phone calls from basic (non-smart) phones. These services have reached over 300,000 low-literate and non-tech-savvy users in Pakistan, India, and West Africa. His work shows that speech-based entertainment services, social networks, and crowd platforms that are accessible over simple and feature mobile phones are powerful motivators to induce self-training and viral spread among under-served populations. Such services can be used as vehicles for large-scale dissemination of information, pulling information in the form of real-time surveys, and performing randomized controlled trials and demographic studies. These services provide target users with visibility, a voice, and digital social identity. His work in ML and Language Technologies focuses on tailoring and democratizing these technologies for the developing world and resource-constrained settings. He developed the first medium vocabulary speech recognition system for Urdu in 2009. He also developed tools, including Urdu text-to-speech, spoken term detection, speech emotion identification, voice biometrics, and audio deepfake detection and techniques, including supervised and unsupervised pruning, to bring large datasets within the computational reach of moderate computing resources. His work on the nexus of ML and Social Good ensures safe and ethical technologies by proactively identifying and mitigating the social impact of potential harms like misinformation and deepfakes, especially among tech-naive and low digital literacy populations. He draws on methods from various disciplines, including mixed methods (qualitative and quantitative), user studies, surveys and focus groups, RCTs, and participant observations. Raza has authored 44 research papers published in peer-reviewed conferences and reputable journals. His novel approach to mass-disseminate information about job opportunities to low-literate mobile phone users in Pakistan received the 2013 best paper award at the ACM CHI Conference on Human Factors in Computing Systems. His work on developing a unique method to disseminate maternal health information to expectant parents in developing countries received an Honorable mention at CHI. His students and RAs have co-authored several of these publications. He has been awarded highly competitive grants by the UNICEF Innovations Fund, GIZ, Facebook/Meta, the Gates Foundation, the National Institutes of Health (NIH), the National Academies of Sciences (NAKFI), Google Inc., and the Higher Education Commission of Pakistan. As a lead PI from Pakistan, he has received grants worth USD 1,574,500 from these prestigious organizations. He has been an Associate Chair for CHI since 2019, the short papers PC chair for ICTD 2017, the open session chair for ICTD 2022, and the Associate Editor for the ACM Journal on Computing and Sustainable Societies (JCSS).

Website: http://aghaaliraza.com

Last updated: Jul 2023