Welcome to my GitHub! I’m an Entry-Level Data Scientist with industry experience at Rolls-Royce Power Systems, passionate about turning data into actionable insights through machine learning, statistical analysis, and scalable data pipelines.

My work spans exploratory data analysis, predictive modeling, time-series analytics, and applied AI systems deployed in real-world environments.

• 📊 Exploratory Data Analysis & Data Wrangling — cleaning, feature engineering, insight discovery
• 📈 Predictive Modeling — classification, regression, ensemble methods
• ⏱️ Time-Series Analysis — segmentation, anomaly detection, forecasting
• 📄 Unstructured Data Analytics — Retrieval-Augmented Generation (RAG), document intelligence

• Built and deployed machine learning pipelines for time-series classification and thermal prediction on real-world sensor data.
• Optimized deep learning models for production, reducing model size from ~29 GB to <1 GB with minimal accuracy loss.
• Accelerated end-to-end inference pipelines using PyTorch GPU acceleration, reducing processing time by ~95%.
• Developed physics-informed ML models to estimate engine node temperatures with ±5–6°C accuracy, reducing reliance on costly simulations.
• Collaborated with cross-functional engineering teams and stakeholders to validate model outputs against domain constraints.

• Built and evaluated ensemble ML models achieving 81% accuracy and 87% F1-score.
• Performed feature engineering and model comparison to identify key churn drivers and derive actionable insights.

• Conducted EDA including data imputation, outlier detection, correlation analysis, and feature distribution analysis.
• Delivered clean, ML-ready datasets for downstream modeling.

• Implemented a RAG pipeline using FAISS and transformer models for semantic retrieval and context-aware answer generation over large PDF documents.

• Compared KNN, Logistic Regression, and Decision Tree models to analyze performance trade-offs in weather prediction tasks.

Languages & Data:
Python · SQL · PostgreSQL · Excel

Data Science & ML:
Pandas · NumPy · Scikit-learn · Matplotlib · PyTorch · TensorFlow

Tools & Platforms:
Git · Jupyter · AWS (Learning)

• Indian Patent (Published): AI-Driven Multi-Factor Authentication Using Hardware Fingerprints
• IEEE Access: Robust Authentication Using Hardware Fingerprints and AI
• Springer: Residual Network Depth vs Accuracy for Crop Disease Classification
• Elsevier (Under Review): Autoencoder-Based Mixed Pixel Correction in Thermal Images

• 📧 Email: sachinsmanoj02@gmail.com
• 🌐 LinkedIn: linkedin.com/in/sachinsm2002

Thanks for visiting! Always open to discussions on data, machine learning, and real-world problem solving 🚀