Arpit Ramesan

AI Generalist & Orchestrator with a talent for building end-to-end AI products, from Generative AI workflows to custom deep learning models. I bridge the gap between advanced research and user-facing software. Driven by a passion for creating more robust systems, I am currently authoring a paper on Neuro-Symbolic AI to explore the intersection of data-driven learning and rule-based logic.

Generative AI Intern - Team Lead at Saspo World Technologies Pvt. Ltd. (2025-09 – 2026-03)

• Spearheaded end-to-end Generative AI systems and creative content production workflows, orchestrating advanced pipelines across multi-model architectures including Google Veo 3.1 and Wan 2.2.
• Directed and managed a 4-member team, standardizing structural engineering procedures, prompt optimization strategies, and system execution practices.
• Engineered tailored system prompts utilizing strict JSON schemas and programmatic prompting protocols, reducing model hallucination and significantly improving semantic coherence across varied model outputs.

Machine Learning Engineer at Incramania Pvt. Ltd. (2023-11 – 2025-09)

• Architected, trained, and deployed a custom cultural Theyyam Detection System using a hybrid CNN+RNN architecture, successfully achieving 79% real-time classification accuracy.
• Designed and integrated production-grade backend engines for multiple machine learning applications utilizing the Flask framework.
• Managed end-to-end computer vision pipelines incorporating TensorFlow, OpenCV, and Roboflow from localized development environments to cloud systems.

B.Tech in Computer Science Engineering – APJ Abdul Kalam Technological University (KTU) (2019-05 – 2023-06)