I am currently a visiting researcher at the CNRS, Sorbonne Université, where I design and study interactive systems that extend human cognition.
My work explores how we can combine human-centered design, AI, and novel interfaces to support sensemaking, reflection, and creativity in complex knowledge work. My research spans interactive machine learning and teaching, AI-augmented interfaces and systems for re-finding, co-writing, and understanding information. I envision, design, build and study systems that champion people's agency to achieve their goals and support them in thinking better.
In addition to academic research, am no stranger to applied work in roles in industry where I designed, managed and delivered products both conceptual and that were shipped.
I am a well-published author in top-tier conferences in human-computer interaction, and a mentor to many graduate students. I hold dozens of patents spanning human-computer interaction, artificial intelligence, interactive machine learning, data visualization, and mixed reality. Many of these patents have informed real-world systems and prototypes at Microsoft and Amazon.
Here you can inquire about some of my research (some N papers and journal articles between 2003 and 2025). Be kind. Save tokens, and the environment. Pay it forward. AI answers can be innacurate. This is no agent. Consume responsibly.
Full list of publications on Google Scholar.
Marquardt, N., Roseway, A., Romat, H., Panda, P., Pahud, M., Ramos, G., Drucker, S., Wilson, A. D., Hinckley, K., Riche, N.
ACM Symposium on User Interface Software and Technology, 2025
ImaginationVellum is a multi-modal canvas for early-stage visual ideation and sketching with generative AI, where the canvas is the prompt. ImaginationVellum employs the entire 2D canvas as an active prompt space, where layout and composition of elements steer generative visual outcomes.
Palani, S., Ramos, G.
arXiv preprint 2508.19517, 2023-25
Orchid is an AI-fueled notebook that seamlessly integrates LLMs into creative workflows with rich context management, bespoke commands, and access to expert personas. Participants using it produced more novel and feasible outcomes with stronger intent alignment, control, and transparency versus baseline.
Tankelevitch, L., Glassman, E. L., He, J., Kittur, A., Lee, M., Palani, S., Sarkar, A., Ramos, G., Rogers, Y., Subramonyam, H.
arXiv preprint 2508.21036, 2025
This report synthesizes outcomes from the CHI 2025 Tools for Thought workshop, examining how generative AI supports and undermines human cognition. It identifies key tensions, risks, and design opportunities for systems that enhance productivity and creativity while preserving user agency and cognitive growth.
Palani, S., Ramos, G.
Proceedings of the 16th Conference on Creativity & Cognition (C&C '24), 2024
How do creative practitioners incorporate generative AI across project stages? Using interviews, videos, and survey reflections on completed projects, we identify perceived roles, challenges, benefits, and interaction patterns, then propose design opportunities for more human-centered creativity support tools.
Yeh, C., Ramos, G.
arXiv preprint 2402.08855, 2023-24
GhostWriter is a collaborative human-AI writing system that introduces agency cues and LLM feedback to better preserve writers' agency and ownership while still improving writing quality. Across two studies, the paper reports that participants retained stronger ownership and preferred workflows that make AI involvement explicit and controllable.
Jörke, M., Sefidgar, Y. S., Massachi, T., Suh, J., Ramos, G.
International Conference on Intelligent User Interfaces, 2023
Pearl is a calendar-based interactive system where users visualize data sources, tag regions of interest, and receive machine-generated insights saved to a reflection journal. In a study with 12 non-expert participants, all gained insights into their workplace wellbeing.
Ramos, G., Rachatasumrit, N., Suh, J., Ng, R., Meek, C.
ACM Transactions on Interactive Intelligent Systems (TiiS), 2022
Guided by sensemaking theory, we built ForSense, a browser extension for accelerating people's online research experience by integrating online research's multiple stages, and providing machine assistance by leveraging recent advances transformer models.
Ramos, G., Meek, C., Simard, P., Suh, J., Ghorashi, S.
Human-Computer Interaction, 2020
We describe interactive machine teaching (IMT) and its potential to simplify the creation of machine-learned models where the human-in-the-loop takes the role of a teacher teaching a machine how to perform a task. IMT processes can enable people to leverage their intrinsic teaching capabilities, and make machine learning methods accessible to subject-matter experts, while creating semantic, debuggable machine learning models.
Dörk, M., Riche, N.H., Ramos, G., Dumais, S.
IEEE Transactions on Visualization and Computer Graphics 18(12), 2012
PivotPaths visually and interactively permits the exploration of faceted information resources, by revealing visual paths that invite the viewer to 'take a stroll' through an information space while encouraging exploration and serendipitous discoveries.
Dragicevic, P., Ramos, G., Bibliowitcz, J., Nowrouzezahrai, D., Balakrishnan, R., Singh, K.
Proceedings of the twenty-sixth annual SIGCHI conference on Human factors in computing systems, 2008
We present a method for browsing videos by directly dragging their content. This method brings the benefits of direct manipulation to an activity typically mediated by widgets and empowers users on space-defined seeking task.
Ramos, G., Cockburn, A., Balakrishnan, R., Beaudouin-Lafon, M.
Proceedings of the SIGCHI conference on Human factors in computing systems, 2007
Pointing lenses help users acquire and select targets by presenting them with an enlarged visual and interaction area. We present and study three pointing lenses for pen-based systems and show that participants find all pointing lenses beneficial.
Ramos, G., Balakrishnan, R.
Proceedings of the SIGCHI conference on Human factors in computing systems, 2007
Pressure marks are pen strokes where the variations in pressure make it possible to indicate both a selection and an action simultaneously. Users perform faster selection-action interactions with pressure marks than with a combination of lassos and pigtails. We present and discuss a number of interaction designs that incorporate pressure marks.
Ramos, G., Robertson, G., Czerwinski, M., Tan, D., Baudisch, P., Hinckley, K., Agrawala, M.
Proceedings of the working conference on Advanced visual interfaces, 2006
We characterize a design space of techniques that facilitate access to occluded graphic content. We also introduce two new tools, Tumbler and Splatter, which represent unexplored areas of the design space and provide a more direct way to engage with overlapping and complex graphic assets.
Ramos, G., Balakrishnan, R.
Proceedings of the 18th annual ACM symposium on User interface software and technology, 2005
Zoom Sliding enables the fluid integrated manipulation of scale and parameter manipulation through the simultaneous control of force and movement input. We introduce and study the concept through controls, prototypes and studies.
Hinckley, K., Ramos, G., Guimbretiere, F., Baudisch, P., Smith, M.
Proceedings of the working conference on Advanced visual interfaces, 2004
Stitching is a new interaction technique that allows users to combine pen-operated mobile devices with wireless networking by using pen gestures that span multiple displays. To stitch, a user starts moving the pen on one screen, crosses over the bezel, and finishes the stroke on the screen of a nearby device.
Ramos, G., Balakrishnan, R.
Proceedings of the 16th annual ACM symposium on User interface software and technology, 2003
We explore interaction and visualization techniques for fluid navigation, segmentation, linking, and annotation of digital videos. These techniques are developed within a prototype leveraging the use of pressure-sensitive digitizer tablets.
Here is a curated list of AI-generated summaries of my work, in the form of podcast conversations.
2026-08-04 · 08:56
A discussion gnerated from a 2022 NeurIPS position paper connecting Jorge Luis Borges' stories — the Aleph, the Library of Babel, the Book of Sand — to the latent space behind image generators like DALL-E 2 and Stable Diffusion, and what that metaphor suggests for designing more human-centered, navigable AI interfaces beyond the prompt box.
based on: paper
Microsoft, 2018-20
Contributed to the study, design, and development of an interactive teaching environment for the creation of machine learning models for classification and entity extraction tasks on text documents.
Amazon, 2015-16
Lead UX and design for an incubation team developing a mixed-reality telepresence system where customers could experience remote locations from home through local guides. Prototyped first versions of the front-end.
Amazon, 2013-16
Lead applied research on future and speculative products. I designed, prototyped, architected, engineered experiences that included, hardware, software, services, mobile, wearables, IoT, TV, AR, NUI, and modern web technologies.
Amazon, 2014-15
Contributed to the concept development, produced interactive prototypes and UX designs for a style assistant consisting of a hardware unit plus software services.
Amazon, 2014
Prototyped first version that processed photos for production, developed decision logic and worked with launch team on a system that algorithmically creates thumbnails that put physical products against a human silhouette. Currently in production displaying 1M+ products.
Microsoft, 2013
Designed among others key visual & interactive aspects of the Photosynth 2 experience. Lead Photosynth's engineering team responsible for delivering the web-viewer & synthing back-end pipeline.
Microsoft, 2011
Designed, prototyped and implemented an experience that blends WWT data with Bing Maps and its StreetSide imagery, along with the creation of a novel timebrowsing interaction technique.