Shankar Koduvayur Ramaswami

About

AI engineer building and deploying deep-learning-based optimization/control systems for petrochemical plants, with a focus on maintaining operational stability under real-world constraints. Core contributor to model and inference design; introduced a stability-focused non-linear objective and sped up second-layer optimization via on-the-fly first-order approximations. Experienced using Kubernetes for end-to-end testing and effective in translating customer expectations into measurable evaluation plots for non-technical stakeholders.

Key Strengths

• Deployed deep-learning optimizer to a petrochemical plant improving operational efficiency and reducing workforce costs
• Bridges traditional APC stability techniques with deep-learning-based control/optimization
• Designed non-linear optimization objective to maintain plant stability under optimization
• Reduced inference time by converting deep learning models to first-order models on the fly (accuracy/speed tradeoff)
• Production-minded orchestration and debugging using Kubernetes (pods, logs, restarts) for end-to-end testing
• Structured AI agent reliability approach: tool-call/parameter validation, adversarial inputs, prompt versioning
• Strong collaboration with non-technical stakeholders using expectation-driven evaluation schemes and visualizations

Experience

Education

Skills

Languages