Data Scientist and ML Engineer with 4+ years of experience building and deploying machine learning models across energy markets, industrial IoT, and NLP/speech domains. Specialized in time-series forecasting, predictive fault detection, stochastic optimization, and deep learning. Currently building real-time BESS bid optimization and forecasting systems at Fluence Energy for AEMO NEM, MISO, CAISO & ERCOT markets.
- 🔭 Currently working on real-time ML price forecasting & BESS bid optimization at Fluence Energy (AEMO NEM, MISO)
- 🌱 Deep expertise in Time-Series Forecasting, LSTM, XGBoost, Stochastic Optimization, Anomaly Detection
- 🤝 Open to Data Science, ML Engineering, and AI/Forecasting roles
- 📫 Reach me at puneethreddy951@gmail.com
- ⚡ Fun fact: My models run every 5 minutes to optimize battery bidding in live electricity markets!
| ⚡ Energy Market Forecasting | Real-time LMP & DART spread prediction for BESS bid optimization (AEMO NEM, MISO, CAISO, ERCOT) — 5-min interval ML pipelines in production |
| 🔋 Battery Fault Detection | Cross-functional analytics on Grid Stack Battery Module sensor data for predictive maintenance (Fluence x NISPERA x FOS) |
| 🌬️ Wind Turbine Health Prediction | Multi-sensor fault prediction & RUL estimation for wind farm asset management using LSTM, Random Forest, Azure IoT + Digital Twins |
| 🎙️ Speech NLP & IVR Automation | STT/TTS pipelines on call recordings, intent/sentiment extraction, automated IVR conversational agents using Azure Cognitive Services |
| ML / Deep Learning | Scikit-learn, XGBoost, LightGBM, LSTM, Random Forest, Gradient Boosting, Anomaly Detection |
|---|---|
| Forecasting & Optimization | Time-Series Forecasting, Feature Engineering, Stochastic Optimization, ARIMA, Prophet, LMP/DART Modeling |
| NLP / Speech | Azure Cognitive Services (STT/TTS), Speech Recognition, Sentiment Analysis, IVR Automation, Dialogflow |
| Data & Analytics | Python, Pandas, NumPy, Matplotlib, Seaborn, Plotly, Jupyter, SQL, EDA, Statistical Modeling |
| Cloud / MLOps | Azure (IoT Hub, Digital Twins, Stream Analytics, Functions, ML, Bot Services), Docker, CI/CD, Azure DevOps |
| Databases | PostgreSQL, MongoDB, Redis, Azure Cosmos DB, Time Series Insights |
| Programming | Python (primary), SQL, JavaScript, TypeScript |
| Energy Domain | AEMO NEM, MISO, CAISO, ERCOT, JEPX, BESS Optimization, Ancillary Services, Grid Stack Analytics |
Feb 2025 – Present
Mosaic BESS Bid Optimization Platform — AEMO NEM, MISO, CAISO, ERCOT
- Built and enhanced production ML price forecasting models for real-time 5-minute interval bidding on Australia's AEMO NEM market, powering live battery dispatch decisions
- Designed time-series pipelines (LSTM, XGBoost) for LMP & DART spread prediction; outputs feed directly into stochastic optimization for BESS charge/discharge strategies
- Led ML model development for MISO market expansion — new feature engineering, PRA revenue modeling, LRZ-level ancillary service co-optimization
- Cross-collaborated with NISPERA (asset management) and FOS (Fluence OS) teams on fault detection analytics for Grid Stack Battery Modules — analyzing cell-level voltage, temperature & SoC telemetry for anomaly detection
Sept 2021 – Jan 2025
Microsoft Client — Wind Turbine Asset Fault Prediction (IoT / Industrial ML)
- Built end-to-end predictive fault detection model for wind farm management using simulated multi-sensor data (vibration, RPM, temperature, power output, pitch angle, hydraulic pressure)
- Full ML lifecycle: Azure IoT Hub ingestion → Stream Analytics → feature engineering (rolling stats, FFT, cross-sensor correlations) → LSTM / RF / GBM training → fault classification & RUL estimation
- Integrated Azure Digital Twins + Time Series Insights for real-time asset state representation and predictive maintenance alerting
eBay Client — Message Studio (Full Stack)
- Built scalable React.js / Next.js / Node.js / PostgreSQL platform serving millions of email template operations for eBay
Feb 2021 – Sept 2021
- Built Speech-to-Text & Text-to-Speech pipelines on real call recordings using Azure Cognitive Services; NLP layer for intent/entity/sentiment extraction enabling automated IVR routing
- Developed Azure Bot Services conversational agents for end-to-end IVR automation with multi-turn dialogue management
- Created voice bot using Asterisk VoIP, AGI, Google Dialogflow & GCP APIs
| Project | Description |
|---|---|
| AgriConnect | Time-series ML price forecasting system trained on 1.36M rows of agricultural market data; ARIMA/LSTM hybrid with cross-validation |
| GramaSuraksha | WiFi CSI sensing platform with ESP32 + LSTM for warehouse intrusion detection & health monitoring from raw radio signals |
| PublicBook | WhatsApp AI agent with Kannada NLP for rural citizens to query government schemes and live market prices |
- 🥇 Excellence Award 2022 & 2023 — Terawe Corporation, for outstanding contributions to client delivery
