Aadya Arora

Final-year undergraduate at IIT Gandhinagar, majoring in Electrical Engineering with a Minor in Artificial Intelligence. My research interests span Computer Vision, Machine Learning, Vision-Language Models, and Generative AI, with a strong focus on healthcare and multimodal perception.

Incoming Data Scientist at Microsoft IDC (2026). Previously worked at IISc Bangalore, University of Bath, IIIT Hyderabad, IIT Hyderabad, and Microsoft IDC on medical imaging, diffusion transformers, document AI, and autonomous driving.

Email  /  Google Scholar  /  GitHub  /  LinkedIn  /  Website

Education

  • B.Tech. Electrical Engineering, IIT Gandhinagar (2022–Present), CGPA: 8.4
  • Senior Secondary, Delhi Public School R.K. Puram (2022), 94%
  • Secondary, Holy Child School, New Delhi (2020), 97%

Publications

Battery Chemistry Recommendation using Machine Learning
Aadya Arora, S. Patil, P. Bhardwaj
IEEE IAS Journal, under review (2025)
Introduced TEMPEST, an LSTM-based framework for internal battery temperature prediction and chemistry recommendation. GitHub
MedFocusCLIP: Improving Few-Shot Classification in Medical Datasets using Pixel-Wise Attention
Aadya Arora, Vinay Namboodiri
ICASSP 2025
Integrated SAM-based pixel attention with CLIP to enable interpretable and robust few-shot medical classification. Paper
WavShadow: Wavelet-Based Shadow Segmentation and Removal
S. Jain*, Aadya Arora*, V. Vekaria, K. Gandhi, S. Raman
ICVGIP 2024
Proposed a wavelet-enhanced pipeline achieving state-of-the-art shadow segmentation and removal. Paper

Work Experience

  • Data Scientist Intern, Microsoft IDC (May–Jul 2025)
    Developed an LLM-driven system to translate natural language instructions into precise PowerPoint formatting commands. Built scalable structured data pipelines, evaluation frameworks, and improved robustness on complex formatting tasks.
  • Incoming Data Scientist, Microsoft IDC (June 2026)

Research Internships

  • IISc Bangalore (Dec 2024 – Apr 2025)
    Layer-wise ablation and analysis of Diffusion Image Transformers (DiTs) for prompt adherence, semantic control, and training-free image editing using attention manipulation.
  • University of Bath (May – Jul 2024)
    Open-vocabulary few-shot referring image segmentation using HIPIE adapters and CLIP, achieving 85.15 mIoU on RefCOCO(+). GitHub
  • IIIT Hyderabad & IIT Hyderabad (Dec 2023 – Jan 2024)
    Analyzed corner cases in autonomous driving datasets (IDD, CODA) using class-agnostic RPNs for anomaly discovery in unstructured Indian traffic.

Selected Projects

  • Thermal-Vision Interpretability (Aug 2025 – Present)
    Analyzing failure modes of Vision-Language Models on thermal imagery using attention heatmaps, patch similarity, and attention prompting techniques.
  • Next Character Predictor (Mar – Apr 2024)
    MLP-based language model trained on literary corpora with architectural and embedding ablations. GitHub
  • Human Activity Recognition (Jan – Feb 2024)
    Time-series classification on UCI-HAR dataset with advanced preprocessing and feature extraction. GitHub

Skills

  • Programming: Python, C, C++
  • Frameworks & Tools: PyTorch, TensorFlow, OpenCV, MATLAB, Azure OpenAI
  • Deep Learning: CNNs, ViT, SAM, CLIP, Diffusion Models, mmDetection, DeepLabv3+

Positions of Responsibility

  • Reviewer, IEEE Signal Processing Letters (2025–Present)
  • Management Coordinator, Professional Development Council, IITGN
  • Teaching Assistant, Machine Learning; Probability & Statistics
  • Coordinator, Career Exposure & Guidance, PDC IITGN
  • Student Guide, Student Support Services

Achievements

  • Microsoft Pre-Placement Offer – Data Scientist
  • Microsoft Research India Academic Summit 2025 (Invited)
  • JEE Advanced 2022 – AIR 9406
  • KVPY (SX) – Extended List
  • State-Level Handball Player (Delhi)