Avatar
Hernán J. Maina hernan.maina@mi.unc.edu.ar
Ph.D. Student in CS
FaMAF · UNC · 🇦🇷
Avatar
Hernán J. Maina
Ph.D. Student in Computer Science
FaMAF · Universidad Nacional de Córdoba · Argentina 🇦🇷

About

Hi! I’m Hernán 👋

I’m a Computer Scientist currently pursuing a Ph.D. at Facultad de Matemática, Astronomía, Física y Computación (FaMAF) · Universidad Nacional de Córdoba (UNC) · Argentina, with a doctoral fellowship from CONICET. I work at the intersection of computer vision and natural language processing. My research, under the supervision of Dra. Luciana Benotti↗, focuses on understanding the challenges of multimodal vision-and-language models to help visually impaired people perform daily tasks and interact with their environment. In particular, I work with visual question answering (VQA) systems, and the detection and recognition of written text in images.

Before that, I completed a Master’s degree with a thesis on real-time recognition of urban bus lines, under the supervision of Dr. Jorge Sánchez↗. This work applied computer vision techniques to improve public transport accessibility for blind and low-vision users.

While my academic background has shaped my problem-solving approach, my interests go beyond research. I’m passionate about developing technology with real-world impact — especially in accessibility and inclusive design — and I’m always open to new challenges and collaborations that push AI out of the lab and into people’s hands. I’m also an enthusiast of robotics and the Internet of Things (IoT), constantly exploring how these technologies can be integrated into intelligent systems that interact with the physical world.

Curious about what doing a Ph.D. in Computer Science is really like — and what it takes to get through it? Head over to the Maybe you’d like to know section. I’ve shared some milestones, side projects, and things I’ve learned along the way.

Publications

Here you can find a selection of my academic work, organized by year. This includes peer-reviewed articles, workshop papers, co-authored research, and academic work associated with degree programs (entries marked with 🎓 highlight milestones from my academic journey).

My work explores the intersection of computer vision, natural language processing, and accessibility — with a focus on how multimodal AI systems can support blind and low-vision individuals in their everyday lives. Some of my work has also examined social biases and stereotypes embedded in language models, and how these may impact the fairness and inclusiveness of AI systems.

2024
2023
2022
2021
2020
2019

Maybe you'd like to know

Here are some highlights and milestones from my academic journey — including presentations, collaborations, courses, grants, and side projects that shaped my Ph.D. experience along the way.

  • December 2024
    • 🎙️ Speaker at the Responsible Artificial Intelligence Workshop, presenting the talk “Accesibilidad en modelos de Visión & Lenguaje (VLMs)”. The event was part of a collaboration between FaMAF · UNC, the CALMS project (Imperial College London), and the HESEIA project (Fundación Vía Libre). [Slides]
  • December 2023
    • 🎓🧪 Final project for the “Automatic Processing of Situated Language in a Visual Environment” postgraduate course, taught by Dra. Luciana Benotti at FaMAF · UNC. Replicated results from prior research papers and extended them to address a specific problem involving grounded language understanding in visual scenes. [Report1] [Report2]
  • November 2023
    • 🎓🛠️ Workshop instructor (seminar accredited) for “Modelos de Lenguaje a tu Medida”, delivered during the 2nd Argentine NLP Workshop, Córdoba, Argentina.

      “This workshop, introduced key concepts in domain adaptation for Large Language Models (LLM). It covered distinctions between pre-training, fine-tuning, and domain adaptation, along with hands-on practice adapting BERT and GPT-2 models to Spanish and English datasets. Participants worked step-by-step using Colab notebooks and shared their results on Hugging Face platform”. [GitHub].

  • September 2023
    • 🎙️🛠️ Workshop instructor at “Modelos de Lenguaje a tu Medida”, part of the 12th Regional Free Software Conference Córdoba, Argentina. [GitHub]
  • December 2022
    • 🏅 Received a Travel Grant to attend KHIPU 2023 in Montevideo, Uruguay.
  • August 2022
    • 🧩 Joined the FAIR Project under the supervision of Dra. Laura Alonso Alemany and Dra. Luciana Benotti, developing tools to detect Biases in Large Language Models. Project led by Fundación Vía Libre, supported by Women at the Table, the A+ Alliance, EthicsTechLab (University of Notre Dame), and Heinrich Böll Foundation.
  • July 2022
    • 🏅 Diversity & Inclusion (D&I) grant recipient to virtually attend to The 2022 Annual Conference of the North American Chapter of the Association for Computational Linguistics (NAACL 2022).
  • December 2021
    • 🎓🧪 Final project for the “Text Mining” postgraduate course, taught by Dra. Laura Alonso Alemany at FaMAF · UNC. Developed a classifier to generate new categories of visual questions using the VizWiz-VQA dataset. [Report] [GitHub]
  • November 2021
    • 🎓 Seminar presenter of “VQA Systems Designed for People with Visual Impairments” Delivered at FaMAF · UNC, Argentina.

      “The seminar provided an overview of Vision-and-Language (V&L) systems and their relevance in accessibility contexts. It explored why the intersection of visual and linguistic modalities is essential for supporting people with visual impairments. A special focus was given to the VizWiz-VQA dataset—created by and for blind users—which presents unique challenges such as blurry or misframed images, lack of visual context, conversational questions, and unanswerable queries.”. [Slides]

    • 🏅 Diversity & Inclusion (D&I) grant recipient to attend to The 2021 Conference on Empirical Methods in Natural Language Processing (EMNLP 2021).
    • 🎙️ CEIMAF Talk — “Inteligencia artificial que habla de lo que ve”. Presented during Computing Day, part of Science Month celebrations. [Video]
  • June 2021
    • 🎓🧪 Final project for the “Parallel Computing” postgraduate course, taught by Nicolás Wolovick at FaMAF · UNC. Developed VizWiz-BERT, a domain-adapted BERT-based model for the VizWiz-VQA dataset, focusing on optimizing training performance through multi-GPU parallelization. [Report]
  • April 2021
    • 🏅 Diversity & Inclusion (D&I) grant recipient for attend to The 16th Conference of the European Chapter of the Association for Computational Linguistics (EACL 2021)
    • 🏅 Awarded a Doctoral Fellowship by CONICET
  • December 2020
    • 🎓🧪 Final project for the “Neural Networks” postgraduate course, taught by Dr. Francisco Tamarit at FaMAF · UNC. Training of Conditional Deep Convolutional Generative Adversarial Networks (cDCGANs) model to automatic image colorization. [Report]
    • 🧪 Completed the Build Basic GANs course on Coursera. [Certification]
  • September 2020
    • 🧪 Completed the Natural Language Processing with Classification and Vector Spaces course on Coursera. [Certification]