Priyanshu Sankhala

I am currently a Master's student in Computational Modeling and Simulation(Informatiks) at TU Dresden, Germany, specializing in Machine Learning and Computer Vision. I also work as an SHK (Social Media Manager) at the Chair of Highly-Parallel VLSI Systems, managing digital outreach for neuromorphic computing research.

Previously, I worked as a Software Developer at Bharat Petroleum Corporation Limited (BPCL), where I built Azure cloud-based data ingestion pipelines processing data from 23,000 retail outlets. Before that, I was a Machine Learning Engineer at Quantrium.ai, focusing on OCR systems, transaction classification, and analytics pipelines.

I received my B.Tech in Electrical Engineering from the National Institute of Technology, Raipur in 2023. During my undergraduate studies, I was selected as a DAAD WISE Scholar for a fully-funded research internship at Goethe University Frankfurt, and also conducted research at IIIT Hyderabad.

Email / CV / Github / LinkedIn

Updates & Milestones

[Oct 2025] Began M.Sc. in Computational Modeling and Simulation at TU Dresden.
[July 2025] Completed the "Certified Cyber Warrior" certification from IIT Madras.
[Sep 2023] Joined BPCL as a Software Developer to build enterprise Azure data pipelines.
[Jan 2023] Promoted to full-time Machine Learning Engineer at Quantrium.ai.
[May 2022] Selected for the prestigious DAAD WISE Scholarship (top 124 in India) for research at Goethe University Frankfurt.
[Dec 2021] Research Intern at IIIT Hyderabad focusing on Graph Neural Networks for drug discovery.

Research & Publications

My research interests lie at the intersection of Reinforcement Learning, Deep Learning, and Computer Vision. I am passionate about developing models in simulation, training there and testing in physical world. Solving real-world problems efficiently.

	Multilingual Hate Speech and Offensive Content Detection using Modified Cross-entropy Loss Arka Mitra, Priyanshu Sankhala FIRE 2021 (ACM) paper Explored the detection of hate speech and offensive content across multiple languages by proposing and utilizing a modified cross-entropy loss function to handle class imbalances effectively.
	Self-Supervised Multimodal Learning DAAD Summer Research under Goethe University Frankfurt May 2022 - July 2022 Applied Grad-CAM to visualize text–image relations in ALBEF. Investigated deep learning representations mapping language and vision.
	Machine Learning Approach to discover Company homepage Priyanshu Sankhala, Naveen Gabriel, Anidhya Bhatnagar code A mapping task pipeline where specific URLs are accurately mapped onto company names utilizing a supervised machine-learning and NLP approach.

Last updated: May 13, 2026