Headline

Research Data Scientist at Arkray| Generative AI, Multimodal AI, Deep Learning, Computer Vision, Emotion Recognition, Thermal Imaging| IIT KGP PhD

About

Aspiring Data Scientist with hands-on experience in several independent projects across Computer Vision, Deep Learning, Machine Learning, and 1D Data Analysis. My work has spanned domains such as Generative Models, Object Recognition, Human-Computer Interaction, Pattern Recognition, Affective Computing, and Robotics. I am passionate about applying data-driven approaches to solve complex problems and motivated to contribute to organizational R&D initiatives that drive innovation and measurable impact. Eager to collaborate on projects that align with company goals while continuing to expand my expertise.

Location

Pune District

AI stack

claude

codex

opencode

Work experience

Lead Data Scientist at ARKRAY Healthcare Pvt. Ltd., Dec 2023 - Present · 2 yrs 8 mos, Pune District, 1. Designed and Improved deep learning pipelines for medical imaging and biomedical signal analysis to support diagnostic applications. 2. Implemented generative models for super‑resolution and style transfer to enhance image quality and augment datasets. 3. Improvised Object Detection model with attention mechanisms and deployed on cloud platforms, and optimized inference. 4. Adapted lightweight models for edge deployment on resource‑constrained devices, improving efficiency through fused operations and input tuning. 5. Developed custom objective functions and signal decomposition methods to improve calibration reliability in biomedical signals. 6. Built time series models for continuous glucose monitoring and LSTM‑based classification of abnormal hemoglobin, successfully deployed to production. 7. Managed technical support processes and coordinated tasks across team members to ensure smooth project execution.

Lead Data Scientist at Sciverse Solutions Pvt. Ltd., Jul 2023 - Dec 2023 · 6 mos, Project 1: Led a team in developing an automated digital microscope for 5-part differential CBC analysis, utilizing computer vision and image processing algorithms to enhance blood cell counting and classification of WBC types. Crafted project proposal, project roadmap, and successfully deployed the microscope in a clinical setting, and currently improvising each step to increase the quality of data and data analysis. Project 2: Designed and implemented an effective low computational cost algorithm to obtain Cycle Threshold or CT values in qRT-PCR sensor data for TB diagnostics. This approach efficiently detects CT values, aiding in determining RIF and INH sensitivity or resistance.

Computer Vision Engineer at Sciverse Solutions Pvt. Ltd., Jul 2023 - Dec 2023 · 6 mos, • Developed lightweight AI models tailored for point-of-care diagnostic devices, enabling real-time clinical support. • Developed end-to-end computer vision pipelines incorporating camera calibration and geometric corrections for precise spatial analysis in blood cell detection, WBC classification, and RT-PCR CT value tracking, achieving an algorithm efficiency of 92%. • Authored detailed government funding proposals for AI startups in healthcare, including comprehensive planning, allocation strategies, and utilization steps for secured funds. • Optimized deep learning models for low-computation environments, enabling deployment on embedded systems and resource-constrained devices.

Research Assistant at Indian Institute of Technology, Kharagpur, Jul 2017 - Jul 2023 · 6 yrs 1 mo, Kharagpur, West Bengal, India, Title: Study of Emotion and Emotional Intensity using Facial Thermal Imagery The research aims to identify the true emotion of individuals and estimate their intensity using thermal imagery. This work involves the use of Machine Learning and Deep Learning concepts to identify the subtle flux of temperature in the subcutaneous layer of our skin that occurs when emotion elicits. The initial pre-processing tasks involves image enhancement and extraction of facial regions which undergo significant changes. Further, this research study includes ways to improve subject independent emotion recognition by learning adaptive thermal features, extracting perfusion data, and leveraging classification using Multi-Graph Convolution Network. These features were further used to obtain the frame-level emotional intensity during emotion elicitation process.

Project Intern at Indian Institute of Technology Kharagpur, Jan 2017 - Jul 2017 · 7 mos, Kharagpur, West Bengal, India, Project: Old age support integrated services (OASIS) Description: Product development, driven by the interplay of research and innovation, where diligent investigation, creative ideation, and technological advancement converge to create new and improved products that cater to customer needs, address challenges, and foster growth.

Technical Support Engineer at Concentrix, Jul 2016 - Dec 2016 · 6 mos, • Provided technical support for network troubleshooting, gateway resolution and security software.

Technical Support Specialist at Concentrix, Jun 2016 - Dec 2016 · 7 mos, On-site

Education

Doctor of Philosophy - PhD, Indian Institute of Technology, Kharagpur

Master of Technology (M.Tech.), Industrial Electronics Technology, National Institute of Technology Rourkela