Gopalakrishnan Arjunan

Client: Chevron

• Developed a machine learning model to predict pipeline maintenance needs based on sensor data, reducing the likelihood of pipeline failures
• Leveraged time-series analysis and anomaly detection algorithms to identify potential defects in pipeline infrastructure
• Resulted in a 20% reduction in maintenance costs, saving the company approximately $2 million annually
• Automated reporting dashboards to provide real-time health metrics of the pipeline system, enabling proactive maintenance strategies
• Designed an optimisation model to balance energy distribution across different geographical regions, reducing energy transportation costs by 15%
• Utilised Python and Gurobi Optimiser to create energy allocation models, saving approximately $1.5 million in logistical costs
• Designed and deployed machine learning models to analyse geological data and predict optimal drilling locations, improving exploration accuracy by 25%
• Resulted in operational cost savings of approximately $10 million annually by reducing unnecessary exploration activities
• Developed an AI-driven logistics optimisation tool for Chevron’s transportation network, reducing fuel costs and delivery times
• Implemented route optimisation algorithms that saved Chevron an estimated $5 million annually in logistical expenses
• Built an AI-powered carbon emission monitoring tool to track and reduce emissions across Chevron's operations, aligning with sustainability goals
• Helped reduce carbon emissions by 20%, saving Chevron approximately $3 million in regulatory costs annually
• Created a predictive maintenance model for oil well equipment using sensor data, minimising unplanned downtime by 30%
• This initiative saved the company approximately $8 million annually in maintenance costs.

Client: Apple

• Enhanced the Siri NLU system by developing an iOS app in Swift to evaluate and test natural language text-to-intent parsing models, improving user interaction simulations
• Streamlined evaluation metrics for natural language models, focusing on user experience-based metrics, resulting in annual savings of $500,000 and earning recognition from Apple Leadership
• Researched and developed deep learning solutions for predictive anomaly detection, forecasting, clustering and correlating millions of rows of time-series data in minutes against unsupervised methods which take hours
• Reduced the training time of deep neural network models by 90% by implementing distributed Tensor-Flow framework on Apple infrastructure
• Facilitated use of machine learning for SREs by developing an automated platform to train and deploy machine learning models on a given stream or set of data, without having to code (similar to Google AutoML)
• Reduced model deployment time by developing a generic machine learning API on Flask that helps to train machine learning models on cloud
• Developed ultra-low memory footprint keyword spotting models with over 50% improvement over production models by minimizing loss-metric mismatch.
• Developed algorithms using Python, C++, and PyTorch while leveraging internal tools to optimize end-to-end pipelines
• Successfully improved Siri’s user experience by bridging gaps between system parses and user interaction
• Delivered measurable results for consumer-facing products through cross-functional collaboration and cutting-edge technologies.

Client: Heineken Ireland Ltd

• Designed and implemented advanced machine learning models to forecast sales demand across Heineken’s product lines, significantly improving inventory management and reducing stockouts by 25%
• Leveraged Python and scikit-learn to develop predictive algorithms based on historical sales data, seasonal patterns, and external factors such as weather conditions
• Delivered approximately $2.5 million in annual cost savings by minimising overstocking and wastage through accurate demand forecasting
• Designed and deployed optimisation algorithms for delivery routes, reducing logistical costs by 20% and improving on-time delivery performance
• Built a dynamic pricing engine, RouteSmart, to adapt prices in real time based on demand forecasts, leading to a 10% increase in revenue
• Combined, these innovations contributed to $4 million in annual cost savings through enhanced supply chain efficiency and improved pricing strategies
• Created a recommendation engine, EngageAI, that personalised marketing campaigns based on customer behaviour, boosting customer retention by 18%
• Automated campaign management using AI solutions, enabling real-time responses to market trends and achieving a 20% increase in customer engagement.

Client: Dow Chemicals

• Developed a command-line interface (CLI) tool to optimize data preprocessing for chemical industry operations
• Engineered to process over 8,000+ datasets, enhancing algorithm interpretability by 67%
• Leveraged Python, object-oriented programming (OOP) principles, pandas, and comprehensive error-handling techniques to enable seamless and efficient data transformation
• Received “Star of the Month” award from Project Manager Neil Bleiberg, IBM