Sohum Sharma

Results-driven Data Scientist and Computer Science student (graduating 2026) with hands-on experience in machine learning, deep learning, and generative AI. Proficient in Python, PyTorch, scikit-learn, and end-to-end ML pipelines spanning data preprocessing, model development, explainability (SHAP, LIME, Grad-CAM), and cloud deployment (AWS EC2, Docker). Demonstrated ability to build multimodal diagnostic and detection systems achieving >88% accuracy.

Seeking a Data Scientist internship to apply research and deployment skills to high-impact, real-world problems.

Machine Learning Engineer - Nimble Vision

(2025-02)

• Trained a custom YOLOv8 object detection model to identify pets, humans, and trucks for smart surveillance systems, achieving 72% mAP accuracy on a custom-labeled dataset.
• Optimized and deployed the model on Samsung S22 series devices using Qualcomm AI Hub, enabling real-time inference at the edge.
• Designed end-to-end pipeline from data collection and annotation through model training, quantization, and on-device deployment.

Backend Engineer - Nimble Vision

(2024-08 - 2024-09)

• Architected a scalable IoT device management system using Django, integrating with LoRaWAN network servers to achieve improved system stability and seamless connectivity.
• Built responsive frontend interfaces (HTML, CSS, JavaScript) for real-time device monitoring dashboards.

Summer Trainee - Oil and Natural Gas Corporation (ONGC)

(2024-06 - 2024-07)

• Automated end-to-end payroll processing using SAP ABAP, eliminating manual data entry and significantly reducing processing errors.
• Engineered automated data retrieval, computation, and PDF pay-slip generation pipeline, improving payroll delivery accuracy and turnaround time.

B.Tech - Computer Science Engineering (CSE) - PES University (2022 - 2026)

CBSE Class XII - Modern School, Barakhamba Road (2019)