Swagat Gudadhe
Summary
Results-driven Data Scientist with over 3 years of experience delivering comprehensive machine learning and analytics solutions, including data collection, feature engineering, model deployment, and ongoing monitoring. Proficient in time-series analysis, anomaly detection, and predictive modeling using Python and SQL, with a proven track record of applying machine learning techniques to address real-world operational challenges. Currently focused on integrating generative AI and large language models into analytical systems to enhance insight generation and improve decision-making processes. Committed to leveraging advanced technologies to drive innovation and enhance business outcomes.
Overview
3
3
years of professional experience
Work History
Data Scientist
ExactSpace Technologies
03.2023 - 03.2026
- Sensor Health Manager
- Architected and deployed an end-to-end Sensor Health Monitoring system for industrial water treatment plants, continuously evaluating ~115K time-series data points/day across 80 sensors per unit, detecting spikes, stuck-at faults, residual drift, and output drift using statistical baselines, persistence logic, and model error distributions.
- Reduced false sensor alerts by ~25% and improved early fault detection, by introducing multi-stage drift validation (3σ Z-score + consecutive window validation + training-period baselines) and separating sensor degradation, model degradation, and process anomalies into independently traceable signals.
- Identified high-risk operating conditions and engineered domain-driven features for ML-based failure prediction models, delivering actionable insights and visualisations that supported proactive maintenance and reduced unexpected downtime by ~10–15%.
- Identified high-risk operating conditions and engineered domain-driven features for ML-based failure prediction models, delivering actionable insights and visualizations that supported proactive maintenance and reduced unexpected downtime by ~10–15%.
- Optimization of Smart Sootblower Module
- Enhanced data-driven sootblower recommendation module using boiler and furnace sensor data for condition-based sootblowing, reducing manual oversight by ~15–20% and minimizing unnecessary operations, improving overall operational efficiency.
- Deployed real-time, rule-driven notification system for advance operational alerts (~90 minutes) for sootblower cycles, adapting alert timing based on furnace temperature and prior events, enabling proactive operator readiness and coordinated operations.
- Backend & API Development (Applied AI / LLMs)
- Built and optimised APIs for a report builder module to aggregate large volumes of sensor data, improving API response time from ~15 ms to ~3–4 ms and enabling faster performance analysis and operational insights.
- Developed backend services to ingest real-time sensor data and manual inputs via JSON-based web and mobile interfaces, implementing dynamic processing and storage logic based on data type
- Contributed to an MCP (Model Context Protocol) based web application integrating local LLMs (Mistral, LLaMa) from Hugging Face to support context-aware information retrieval using locally hosted models.
Education
MBA - Business Analytics
University of Pune
Pune 04.2001 -
Diploma - Artificial Intelligence and Machine Learning
University of Hyderabad
01-2022
Skills
- Machine learning
- Time series analysis
- Anomaly detection
- SQL proficiency
- SQL databases
- Python programming
- Data visualization tools
- Data presentation
- Statistical analysis
- Version control with Git
Timeline
Data Scientist
ExactSpace Technologies
03.2023 - 03.2026
MBA - Business Analytics
University of Pune
04.2001 -
Diploma - Artificial Intelligence and Machine Learning
University of Hyderabad