Embedded Files
As a Data Scientist in a SaaS company serving the energy sector, I led the development of advanced AI-driven solutions for energy portfolio management, forecasting, and trading. My work spanned the full machine learning lifecycle—from designing credit risk models using financial and operational data to implementing LSTM-based pipelines for short-term energy demand forecasting. I contributed to strategic trading systems, including real-time BESS control algorithms and mean-variance portfolio hedging strategies powered by ML-based volatility prediction. Collaborating within agile DevOps teams, I also delivered tools for risk analysis and portfolio optimization, supporting over 120 enterprise clients across international markets. My expertise lies in turning complex energy data into intelligent software solutions that drive operational efficiency, risk mitigation, and market competitiveness.
Portfolio Analysis and Risk Management
Development of Comprehensive Company-Wide Risk Management and Monitoring Tools
Designed and implemented a system to streamline the assessment and mitigation of trading transaction risks aligned with audit and compliance requirements.
Established robust monitoring mechanisms, including configurable risk limits for procurement staff, portfolios, products, and counterparties.
Enabled proactive risk management by detecting and responding to limit violations through automated actions such as multi-channel alerts, trade execution blocks, and other strategic controls to safeguard trading operations.
Monitored and optimized procurement segments through ongoing economic performance evaluation, improving procurement efficiency and cost-effectiveness.
Developed visual tools for the decomposition of open positions into tradable product structures, enhancing transparency and supporting data-driven trading strategies.
Forecasting
Led the design and theoretical implementation of LSTM (Long Short-Term Memory) pipelines to forecast energy consumption for industrial customers, significantly outperforming traditional models such as multivariate adaptive regression splines and random forests. This initiative delivered substantial improvements in forecast accuracy, enabling energy suppliers to better tailor contracts and agreements with industrial clients. It also supported more effective short-term energy portfolio balancing—across near real-time, day-ahead, and intraday trading horizons—driving smarter procurement and trading decisions.
Option Pricing
In this project, I developed and implemented sophisticated pricing models for both European and American style option contracts, integrating them seamlessly into the portfolio management pipeline. These models played a critical role in accurately evaluating options as part of a comprehensive risk management framework, enabling more precise assessment and mitigation of financial risks associated with energy trading portfolios.
By collaborating closely with cross-functional teams—including risk analysts, traders, and data scientists—I ensured that the option pricing tools aligned with broader portfolio optimization and trading strategies. Continuous monitoring and iterative fine-tuning of the models enhanced the portfolio’s risk-return balance, supporting proactive decision-making under market uncertainty. This approach not only improved the precision of risk quantification but also empowered energy suppliers and traders to optimize hedging strategies, maximize returns, and maintain compliance with regulatory standards in a highly volatile market environment.
Value at risk
I led the development and implementation of a robust Value at Risk (VaR) framework to assess the risk exposure of supplier customers’ open portfolio positions in volatile energy markets. Recognizing VaR as a vital tool for managing market risk, I implemented multiple calculation methods—historical simulation, variance-covariance, Monte Carlo simulation, and exponentially-weighted moving averages (EWMA), to provide a well-rounded, adaptive risk assessment.
Each approach was tailored to reflect the portfolio’s structure, capturing both linear and non-linear risk factors across physical assets and derivatives. The models enabled suppliers to quantify potential losses, strengthen hedging strategies, and support risk-informed decisions across day-ahead, intraday, and longer-term trading. This framework became an integral part of the portfolio management process, aligning risk oversight with strategic trading objectives.
Value_at_Risk__A_Brief_Review.pdfTrading Pipeline
I collaborated with a cross-functional and innovative team spanning Europe and New Zealand. We developed and conceptualized a trading algorithm integrated into an operational pipeline. Within our collaborative team, the trading algorithm took on the crucial task of discerning bid and offer signals based on forecasted prices. Expanding beyond its identification role, the operational pipeline further facilitated the transmission of these signals to system operators in pre-agreement forms, awaiting confirmation. Once confirmation was received, the Battery Energy Storage System (BESS) was prompted through an IoT hub to engage in charging or discharging activities, dynamically tapping into energy sources from both wholesale and reserve platforms. This integrated process significantly heightened the ability of the customer to patch and dispatch energy.
Page updated
Google Sites
Report abuse