Jul 2021 – Jun 2022 · 11 mos

• Lead a team of 11 students to deliver solutions for a live consulting project from Curelink. Market expansion strategy, customer acquisition, and market research were covered under the scope of the project.

Jun 2021 – Apr 2022 · 10 mos

• Designed a novel algorithm to classify skin lesions into Cancerous and Non-Cancerous classes using Multi-Modal Data, and published work on the same in ICPR 2022.
• Image feature-maps obtained from CNNs when fused with the metadata using our algorithm improved accuracy on the SoTA by 11%.
• A separate work done on incorporating trainable visual attention fused with patient metadata was accepted at MICCAI workshop, ML-CDS 2022.
• Worked under the supervision of Prof. Balasubramanian Raman, HoD, CSE Dept, IIT Roorkee.

Sep 2020 – Nov 2020 · 2 mos

• Worked on a consulting project in association with ShARE for nanotech startup Log9 Materials.
• Analyzed the Activated Carbon market in India based on types, pricing and industrial applications.
• Conducted research and interviews to provide insights on activated carbon products of 20 + companies, with a focus on HVAC filters.
• Connected Log9 with a potential client- Blueair.

Jun 2020 – Jul 2020 · 1 mo

• Used deep learning for deblurring of invoice images to increase OCR accuracy for claims automation process.
• Created an artificial dataset of 5600 text images by cropping pdfs of e-books.
• Used CycleGAN and SRResNet models. SRResNet after training on the synthetic dataset and transfer learning on real invoices enhanced the readability of test invoices.