• Atharva Jindam
• Ankit Javeri

Kubernetes-based cloud-native microservices frequently experience issues like resource contention, latency spikes, and pod crashes. Adaptivity is lacking in conventional healing methods (restart, Horizontal Pod Autoscaler). In this paper, a self-healing agent based on Reinforcement Learning (RL) and Proximal Policy Optimization (PPO) is proposed. The agent keeps an eye on telemetry, chooses remedial measures, and reduces downtime and Service Level Objective (SLO) violations. In comparison to baselines, a prototype utilizing Kubernetes, Prometheus, Chaos Mesh, and PyTorch demonstrates that PPO lowers Mean Time to Recovery (MTTR) by approximately 35% and SLO violations by approximately 30%.

Reinforcement Learning, Proximal Policy Optimization (PPO), Kubernetes, Self-Healing Systems, Microservices Architecture, Cloud Resilience, Adaptive Systems, Fault Tolerance, Service Level Objectives (SLO), Chaos Engineering

"PPO-BASED REINFORCEMENT LEARNING FOR SELF-HEALING IN KUBERNETES", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 9, page no.d515-d518, September-2025, Available :http://www.jetir.org/papers/JETIR2509369.pdf

"PPO-BASED REINFORCEMENT LEARNING FOR SELF-HEALING IN KUBERNETES", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 9, page no. ppd515-d518, September-2025, Available at : http://www.jetir.org/papers/JETIR2509369.pdf