Navodita Mathur

Machine learning Engineer, Applied AI Engineer
Location: Jersey City, NJ
Contact: +1(412) 954-7877
Email: navoditamathur1998@gmail.com

About Me

I am a Machine Learning Engineer building multimodal AI systems for life sciences and regulated domains, with a strong foundation in computer vision, applied machine learning, and production-oriented software development.

I bring ~3.5 years of software development experience and hands-on work designing, training, and deploying ML pipelines that operate on complex, real-world data — including medical imaging, clinical records, and large-scale geospatial imagery. I am particularly interested in precision medicine, healthcare AI, geospatial AI and applied ML systems that integrate heterogeneous data sources to support real-world decision-making.

🧬 Healthcare & Life Sciences ML

Multimodal Clinical Decision Support
Built an agentic AI system integrating EHR data, medical imaging, and knowledge graphs to support disease prediction and clinical query resolution.
Focus areas included data normalization, retrieval pipelines, model evaluation, and human-in-the-loop workflows.

Medical Image Segmentation (3D MRI)
Developed and optimized Swin-UNETR and attention-based architectures for 3D brain tumor segmentation.
Worked on model training, evaluation (Dice, sensitivity), and robustness to domain variability in medical imaging data.

🧠 Large-Scale Computer Vision & Imaging

Remote Sensing for Environmental Monitoring
Designed end-to-end ML pipelines using Sentinel-2 satellite imagery for deforestation detection and vegetation analysis.</b> Work included preprocessing, feature extraction, segmentation modeling, and deployment-ready inference workflows.</b>

Infrastructure Damage Detection
Implemented wavelet-based U-Net architectures with attention mechanisms for crack segmentation, emphasizing texture preservation and evaluation metrics for fine-grained image segmentation tasks.

🔬 Applied Research & Prototyping

I use applied research to prototype and validate ML approaches before production use. This includes experimenting with model architectures, evaluation strategies, and domain-specific constraints in medical imaging and large-scale vision tasks.

Selected projects focus on:
attention-based segmentation models
transformer architectures for imaging
multimodal representation learning
evaluation under noisy and imbalanced data conditions

🧾 Publications & Technical Contributions

Paper accepted — ICCV CV4E Workshop
Peer-reviewed paper accepted at the Computer Vision for the Environment (CV4E) workshop, International Conference on Computer Vision (ICCV).

Poster Presentation — ICCV CV4E Workshop
Presented work on geographical priors for fine-grained Butterfly Classification

Paper accepted — AAAI 2026 KGML Bridge
Accepted paper at the Knowledge-guided Machine Learning (KGML) Bridge at AAAI 2026, focusing on Knowledge Graph-Guided Multitask Learning for Bird Species and Trait Recognition

Early Career Talk — AAAI 2026 KGML Bridge
Invited early-career presentation on applied ML and knowledge graph integration for real-world systems.

(Links available upon request / see linked repositories for implementations.)

⚙️ Technical Focus

Languages & Frameworks: Python, PyTorch, FastAPI
Machine Learning: Computer Vision, Multimodal ML, Segmentation, Representation Learning
Cloud & Systems: AWS, Docker, CI/CD-aware development
Data & Imaging: Medical Imaging, Satellite Imagery, Structured & Unstructured Data
Engineering Practices: Modular pipelines, evaluation-driven development, production-minded ML

📌 What I’m Looking For

I am currently seeking Machine Learning Engineer / AI Engineer roles in life sciences, healthcare, or applied ML teams working with complex, real-world data in regulated or high-impact domains.