I build machine learning systems end-to-end — from problem formulation and experimentation to deployment, monitoring, and evaluation.

My background combines 4 years of industrial engineering with a Master’s in Data Sciences & Business Analytics at CentraleSupélec & ESSEC. That journey shaped how I approach ML: understand the system first, establish strong baselines, experiment rigorously, and add complexity only when the evidence supports it.

My recent work spans time-series forecasting, energy systems, deep learning, recommendation systems, LLM applications, and production ML.

• Machine Learning & Deep Learning — robust modelling, evaluation, and experimentation
• Time Series & Forecasting — temporal validation, non-stationarity, energy and operational systems
• Scientific ML & Energy Systems — data-driven modelling of physical and industrial systems
• LLM & Agentic Systems — RAG, agents, orchestration, evaluation, and tool use
• Production ML — reproducible pipelines, monitoring, experiment tracking, and deployment

I care about more than getting a good metric on a test set.

• Start with simple, meaningful baselines
• Use leakage-safe and out-of-time validation
• Design controlled experiments and ablation studies
• Treat negative results as useful experimental evidence
• Evaluate models against the decision they support, not only a loss function
• Prioritize robustness, reproducibility, and monitoring
• Add model complexity when experiments justify it

I also build systems across:

• LLM / Agentic AI — RAG, multi-agent workflows, MCP and structured evaluation
• Recommendation & Ranking — retrieval, learning-to-rank and recommender architectures
• Computer Vision — image translation, detection and segmentation
• ML Engineering — APIs, Docker, CI/CD, orchestration, monitoring and experiment tracking

I’m particularly interested in problems where machine learning interacts with complex real-world systems — from industrial forecasting and energy systems to scientific ML and modern AI applications.

Current areas I’m exploring include:

• Scientific Machine Learning and ML for physical systems
• Probabilistic and uncertainty-aware modelling
• Reliable time-series forecasting
• LLM agents and tool-using systems
• Evaluation and monitoring of production ML systems

Languages: Python · SQL · C++ · R

Machine Learning: PyTorch · TensorFlow · Scikit-learn · LightGBM · XGBoost

LLM / AI Engineering LangChain · LangGraph · RAG · MCP · Vector Search

ML Engineering & Data MLflow · Airflow · Docker · FastAPI · Git · CI/CD · Grafana · InfluxDB

Cloud AWS · Google Cloud Platform

📬 sabariravi29597@gmail.com
🔗 https://www.linkedin.com/in/sabarishravim/

From building car components to building ML systems — still solving systems, just at a different scale.