Solomon Khess

Developed LSTM and XGBoost forecasting models for Superheater Steam Temperature Prediction using high-frequency process sensor data, improving prediction robustness under non-stationary operating conditions.

Applied Hurst Exponent

Analysis, Chaos Theory, and Nonlinear Dynamic Analysis to identify long-memory behaviour and extract complex temporal features from industrial process signals.

Performed Dynamic Feature

Selection using t-statistics, AIC, and mRMR; incorporated Recurrence Quantification Analysis (RQA) metrics to capture system stability and nonlinear dynamics.

ML Research Intern at National Thermal Power Corporation Ltd (2025-06 – 2025-07)

• Developed LSTM and XGBoost forecasting models for Superheater Steam Temperature Prediction using high-frequency process sensor data, improving prediction robustness under non-stationary operating conditions.
• Applied Hurst Exponent Analysis, Chaos Theory, and Nonlinear Dynamic Analysis to identify long-memory behaviour and extract complex temporal features from industrial process signals.
• Performed Dynamic Feature Selection using t-statistics, AIC, and mRMR; incorporated Recurrence Quantification Analysis (RQA) metrics to capture system stability and nonlinear dynamics.
• Engineered lag-based temporal feature pipelines enabling sequence-aware learning and improved forecasting accuracy on noisy thermal power plant data.

ML Research Intern at Shri Mahavir Ferro Alloys Pvt Ltd (2025-05 – 2025-05)

• Developed a predictive ML pipeline using time-series analysis to forecast Rotary Kiln accretion via surface temperature monitoring, mitigating unplanned shutdowns.
• Performed feature engineering on kiln thermal profiles to identify early-stage buildup patterns, delivering data-driven strategies that enhanced kiln stability and runtime.

B.Tech–M.Tech – Indian Institute of Technology, Bhubaneswar (2021)

Class XII – Ascent Junior College (2020)

Class X (CBSE) – Delhi Public School (2018)