Looking for a job

Sushmita Upadhyay

@sushmitaupadhyay

Message

Machine Learning and AI Engineer at Aiverbalyze Technologies, adapting Qwen2.5-Omni and optimizing Gemma models.

India

Message

At Aiverbalyze Technologies Pvt. Ltd., I work with multimodal and speech AI models, including Qwen2.5-Omni and OmniVoice. I contributed to adapting Qwen2.5-Omni and expanding its vocabulary, and explored optimization and pruning for Gemma-based models.

My PhD research at BIT Mesra focused on attention-based visual question answering. I developed models for vision-language understanding, trained and evaluated multimodal models on HPC systems, and optimized lightweight VQA models for the NVIDIA Jetson Orin Nano.

At Comofi Medtech, I implemented a Python pose-estimation algorithm for medical imaging and worked with CT, DRR, C-arm, and multi-camera data. My earlier research included developing a lightweight gait-classification model and deploying it on an embedded device.

Experience

Work history, roles, and key accomplishments

Current

Machine Learning and AI Intern

Current

Aiverbalyze Technologies Pvt. Ltd.

Jun 2026 - Present (4 months)

Worked with multimodal and speech AI models, including Qwen2.5-Omni and OmniVoice. Contributed to model adaptation and vocabulary expansion for Qwen2.5-Omni, and explored model optimization and pruning for Gemma-based models.

Multimodal AI Speech AI Qwen2.5 Omni OmniVoice Model Optimization Pruning Gemma

Junior Research Fellow

BIT Mesra

Nov 2022 - Oct 2023 (11 months)

Analyzed gait data to identify Freezing of Gait (FOG) and Non-FOG patterns in Alzheimer's patients. Developed a lightweight deep learning model for gait classification and deployed it on an embedded device, and developed a prototype for real-time gait data collection and FOG detection.

Deep Learning Gait Analysis Embedded Systems IoT Healthcare Model Deployment

Computer Vision Engineer

Comofi Medtech

Jan 2021 - Sep 2021 (8 months)

Implemented a pose estimation algorithm in Python for medical imaging applications. Worked with CT, DRR, C-arm, and multi-camera imaging data for pose analysis, and contributed to system documentation, workflow design, and project reporting.

Python Pose Estimation Medical Imaging DRR C Arm Multi Camera Documentation

Associate System Engineer

IBM India Pvt Ltd

Aug 2014 - Oct 2017 (3 years 2 months)

Worked as an Associate System Engineer on Project-Birla Sunlife Insurance in Mumbai and as a Graduate Hire in Kolkata. Responsibilities included system engineering and support for insurance project.

Education

Degrees, certifications, and relevant coursework

Birla Institute of Technology, Mesra

PhD, Multimedia Processing and Deep Learning

2021 -

PhD in Multimedia Processing and Deep Learning, focusing on attention-based Visual Question Answering systems. Developed and optimized models for edge deployment.