Sankalp Arora

Research Associate @Edinburgh Centre For Robotics

Edinburgh, GB

Engineered a supervised ML pipeline using PyTorch to train intelligent gameplay agents for Super Mario Bros on state-action pairs, achieving 50% faster convergence and better generalization than baseline RL models• Developed a modular MLOps workflow to automate data ingestion, model retraining, and deployment, reducing manual deployment time by 60% using Streamlit and Flask APIs for real-time inference• Integrated LLM-based agents (OpenAI GPT-4 via LangChain) to simulate human-like decision paths, augment training data, and generate KPI-aligned sequences for evaluation (e.g, completion rate, action entropy)• Deployed containerized inference services using Docker, with reproducible training environments, versioned model tracking, and exploratory infrastructure setup via AWS S3 + EC2 for compute

National Institute of Technology, Kurukshetra, Haryana

Bachelor's degree, Electrical and Electronics Engineering

Children senior secondary school

Intermediate/+2, Science and maths

Heriot-Watt University

Masters, Mechatronics, Robotics, and Automation Engineering