About Me
I am an AI Research Engineer at MBZUAI, working with Prof. Salman Khan on multimodal foundation models and agentic AI for climate analysis (Climate Co-Pilot). Previously, I was with the BioMedIA Lab at MBZUAI under Prof. Mohammad Yaqub, first as an M.Sc. student in Computer Vision and then as a Graduate Research Assistant. My work there covered echocardiography, fetal ultrasound, and brain tumor segmentation, including FetalCLIP, a vision-language foundation model for fetal ultrasound.
In 2025, I worked as a Senior Data Scientist at Miral Experiences in Abu Dhabi, transforming park data into business insights and recommendations, from personalized recommendations for visitors across four parks and price recommendation systems for retail and F&B to advanced ad hoc analytics for business leaders. Across research and industry, I enjoy taking a problem from careful data curation through modeling to a solution that people can use and build upon.
I am always happy to hear from researchers and teams working on related problems, so please feel free to reach out by email. To give my current commitments the focus they deserve, I am unfortunately not able to take on any part-time roles at the moment.
Research interests: multimodal models, foundation models, agentic AI, AI for healthcare, AI for climate, and applied AI & ML.
News
- Oct 2026 Preprint New preprint: EarthAlign, with the dataset, code, and model weights publicly released.
- Jun 2026 Publication 1 paper published in npj Digital Medicine: FetalCLIP.
- Jun 2026 Preprint New preprint: TerraBench.
- Mar 2026 Preprint New preprint: DARK.
- Feb 2026 Award Received the Impactful Research Award for FetalCLIP at MBZUAI.
- Feb 2026 Career Joined MBZUAI as an AI Research Engineer, working with Prof. Salman Khan.
- Nov 2025 Publication 1 paper accepted at WACV 2026: DuPLUS.
- Apr 2025 Career Joined Miral Experiences, Abu Dhabi, as a Senior Data Scientist.
- Feb 2025 Publication 1 paper accepted at CVPR 2025 (Highlight): All Languages Matter (ALM-Bench).
- Oct 2024 Award CoReEcho presented as an oral at MICCAI 2024 and nominated for the Best Paper and Best Young Scientist Paper awards.
- Oct 2024 Challenge 3rd place in BraTS 2024 Pediatrics & Sub-Saharan Africa at MICCAI 2024: Optimizing Brain Tumor Segmentation with MedNeXt.
- Oct 2024 Publication 1 paper accepted at WACV 2025: ConDiSR.
- Sep 2024 Publication 1 paper published in Ultrasound in Medicine & Biology: SimLVSeg.
- Jun 2024 Career Joined the BioMedIA Lab at MBZUAI as a Graduate Research Assistant.
- Jun 2024 Education Graduated with an M.Sc. in Computer Vision from MBZUAI (GPA 4.00/4.00).
- May 2024 Publication 2 papers accepted at MICCAI 2024: CoReEcho and SurvRNC.
- May 2024 Challenge 2nd place in the BraTS 2024 GoAT challenge at ISBI 2024: On Enhancing Brain Tumor Segmentation Across Diverse Populations with Convolutional Neural Networks.
- Oct 2023 Challenge 3rd place in BraTS 2023 Adult Glioma at MICCAI 2023: Advanced Tumor Segmentation in Medical Imaging.
- Oct 2023 Publication 1 paper published at MLMI 2023: Exploring the Transfer Learning Capabilities of CLIP in Domain Generalization for Diabetic Retinopathy.
- May 2023 Career Started a graduate internship at LabibAI, Abu Dhabi.
- Aug 2022 Education Joined MBZUAI for an M.Sc. in Computer Vision on a full scholarship.
Selected Publications
The full list of publications is available on Google Scholar.

FetalCLIP: A Visual-Language Foundation Model for Fetal Ultrasound Image Analysis
F. A. Maani*, N. Saeed*, T. Saleem, Z. Farooq, H. Alasmawi, W. Diehl, A. Mohammad, G. Waring, S. Valappil, L. Bricker, M. Yaqub
npj Digital Medicine, 2026
- A vision-language foundation model for fetal ultrasound, pretrained on over 210K image–caption pairs and evaluated on plane classification, gestational age estimation, congenital heart defect detection, and segmentation.

EarthAlign: Learning Language-Aligned Representations of Earth System Data
F. A. Maani, N. Saeed, J. Ma, D. T. Nguyen, Y. Wehbe, D. Watson-Parris, S. Khan
arXiv preprint, 2026
arXiv | Project | Code | Dataset & Weights
- A 553K-sample dataset pairing natural-event and normal-weather records with ERA5 meteorological data and grounded captions, plus MeteoLIP, a dual-encoder model for meteorology–language alignment.

N. Saeed, A. Hanif, F. A. Maani, H. Alasmawi, M. Yaqub
arXiv preprint, 2026
- A contrastive distillation method that compresses FetalCLIP into MobileFetalCLIP, with a 26× smaller visual encoder that runs in 1.6 ms on an iPhone 16 Pro while matching or exceeding its teacher on zero-shot benchmarks.

CoReEcho: Continuous Representation Learning for 2D+time Echocardiography Analysis
F. A. Maani, N. Saeed, A. Matsun, M. Yaqub
Medical Image Computing and Computer Assisted Intervention (MICCAI), 2024
- A training framework that learns continuous representations for direct ejection fraction regression, reaching state-of-the-art results on EchoNet-Dynamic and transferring better to smaller echocardiography datasets.

F. A. Maani, A. Ukaye, N. Saadi, N. Saeed, M. Yaqub
Ultrasound in Medicine & Biology, 2024
- Self-supervised temporal masking followed by weakly supervised training lets video networks segment the left ventricle from sparsely annotated echocardiograms, reaching 93.32% Dice on EchoNet-Dynamic.

F. A. Maani*, A. U. R. Hashmi*, M. Aljuboory, N. Saeed, I. Sobirov, M. Yaqub
BrainLes Workshop @ MICCAI, 2023
- An ensemble of SegResNet and MedNeXt with lesion-wise post-processing, placing 3rd in the BraTS 2023 Adult Glioma challenge.
Experience
Research
- Working on Climate Co-Pilot, which brings multimodal foundation models and agentic AI to climate analysis; a collaboration with ADIA Lab and UC San Diego, funded by ADIA Lab.
- Released EarthAlign, a 553K-sample dataset and models for aligning Earth system data with language.
- Developed FetalCLIP, a vision-language foundation model for fetal ultrasound; a collaboration with Corniche Hospital in Abu Dhabi, partially funded by GE HealthCare.
- Contributed to medical image segmentation research.
- Mentored a team that achieved 3rd place in BraTS 2024 Pediatrics & Sub-Saharan Africa at MICCAI 2024 (paper).
- Developed state estimation for an autonomous vehicle, including localization with a virtual sensing system and when GNSS information is missing.
- Developed longitudinal and lateral control, from CARLA simulation to deployment on a real golf cart.
Industry
- Built a personalized next-action recommender for visitors across four parks, and price recommendation systems for retail and F&B.
- Delivered advanced ad hoc analytics for requests from business leaders.
- Tools: Databricks, Snowflake, and Power BI.
- Built TAMM-GPT, a pilot chatbot that answers questions about UAE government services.
- Developed an inspection system for presence detection, object counting, and barcode and data-matrix reading, implemented at Toyota Motor Manufacturing Indonesia (TMMIN).
- Tumor segmentation in whole-slide images and Fasciola egg detection in microscopic images.
- Developed a low-cost, internet-based automation system using Arduino for the Press & Welding Production Department.
Honors and Awards
Education
- GPA 4.00/4.00, full scholarship.
- Thesis: On Enabling Video Networks for Reliable 2D+Time Echocardiography Analysis.
- GPA 3.96/4.00, cum laude.
- Thesis: Implementation of Longitudinal and Lateral Control on Autonomous Vehicle.