David Nnamdi, SPE, is a senior data scientist at Intuit. He has expertise in statistical analysis, machine learning (ML), distributed computing, and convex optimization, leveraging analytics to drive impactful business decisions. Passionate about deep learning, he focuses on building adaptable, domain-specific solutions.
He holds dual master’s degrees in petroleum engineering, data science and analytics, and a bachelor’s degree in petroleum and gas engineering.
In this interview, I spoke with Nnamdi about his work as a data scientist and engineer, his development of Sequestrix, an open-source CO2-transport-network optimization tool, and where he sees data science and AI’s role in the future of sustainable energy.
Abdulmalik Ajibade (AA): You started your career as a reservoir engineer at Niger Delta Exploration & Production plc. Can you tell us about what first sparked your interest in the energy sector and your early work there, including your contributions?
David Nnamdi (DN): My interest in the energy sector began in my early teenage years, as oil and gas is one of the most important sectors in my home country, Nigeria. I earned my undergraduate degree in petroleum engineering from the University of Lagos and immediately started working as a reservoir engineer at NDPR (now Aradel Energy). The company pioneered the development of marginal oil fields in Nigeria, where efficiency was crucial.
As a member of a small team, my role was broad and required an end-to-end understanding of reservoir engineering, from geology and well-log interpretation to production forecasting and reservoir surveillance. Some of my key accomplishments included creating innovative production-forecasting tools, being part of the team that delivered the company's first horizontal wells, and co-designing transient tests for the company's first offshore exploration well.
AA: During your time as a reservoir engineer, you also worked as a software developer and data analyst, co-developing a dashboard for well- performance monitoring and creating internal scripts for pressure transient analysis (PTA). What motivated you to start integrating technology and data analysis into traditional petroleum engineering roles so early in your career?
DN: With expectations to deliver high-impact work, I looked for ways to improve speed, accuracy, and reliability. At the time, production forecasting and PTA workflows were largely manual. Two early mentors inspired me to learn Python and data science, and I quickly realized I could automate much of the manual data work. I built a Python tool for large-scale production forecasting, followed by PTA scripts and well-performance monitoring dashboards. These tools boosted my productivity tenfold and enabled faster, better asset management decisions by blending petroleum engineering with programming and analytics.
AA: You later pursued two master's degrees simultaneously at the University of Oklahoma: one in petroleum engineering and another in data science and analytics. What led you to make the decision to pursue both of these fields at the same time?
DN: I pursued both degrees simultaneously because I saw the future of engineering was moving toward data-driven decision-making. The petroleum engineering degree gave me the domain expertise to understand complex subsurface systems, while the data science degree equipped me with the analytical and computational skills to model and optimize those systems. Additionally, I had already applied data science in my reservoir engineering work, including publishing papers on using neural networks for production forecasting and proxy reservoir models, and I wanted to deepen that path.
AA: At the University of Oklahoma, you were a research scientist and teaching assistant. Your work included using clustering algorithms to aggregate CO2-injection sinks and investigating CO2-capture technologies. How did your academic research connect your petroleum engineering background with your growing expertise in data science?
DN: My research bridged petroleum engineering and data science by applying computational methods to carbon capture and storage (CCS). CCS projects require identifying suitable sinks, matching them to CO2 sources, designing pipeline networks, and assessing economic viability.
I used clustering algorithms to group 245 injection sites into 21 clusters, cutting computational complexity for network optimization. I also developed a multicriteria decision framework to evaluate capture sites and technologies, incorporating thermodynamics, economics, and life-cycle emissions. This work showed how modern data science tools could streamline CCS planning and investment decisions
AA: Speaking of your research, you developed Sequestrix, an open-source CO2-transport-network optimization tool. Can you tell us more about Sequestrix and how it helps accelerate economic analysis for CCS projects by combining graph-based algorithms and optimization?
DN: Building on my CCS research, I developed Sequestrix, an optimization tool that models CO2-transport networks as graphs. Using mixed-integer linear programming, it identifies the most cost-effective combination of pipeline routes and capacities while factoring in existing infrastructure and operational constraints—its key innovation. This allows project planners to run rapid "what if" analyses, assess investment scenarios, and accelerate feasibility studies.
By making Sequestrix open-source, I aimed to democratize access to these advanced optimization capabilities and help governments, researchers, and the industry move CCS projects from concept to implementation more quickly.
AA: After your master's degrees, you transitioned into roles as a data scientist at Pioneer Natural Resources Company and later at LinkedIn and Intuit. Can you walk us through that transition? What was the biggest challenge, and what was the most surprising similarity between the two fields?
DN: My transition was driven by the realization that my data science skills could be applied across many domains, not just reservoir engineering. The biggest challenge was rewiring my mind to focus on technology-based problems rather than energy-based ones. I found many similarities across industries; for example, both the energy and technology fields collect high-fidelity data, from well-production sensors to real-time user-engagement data.
The statistical and ML methods used to analyze this data are largely the same, differing only in the specific problem you are solving, like predicting well-integrity issues versus predicting user engagement or optimizing water allocation in the Permian Basin versus ad spend in digital marketing.
AA: While at Pioneer Natural Resources, you co-developed and operationalized optimization tools like WaterWise and SandWise, which significantly enhanced a $2-billion operation. Can you explain the technical side of this a bit more? How did you use Gurobi-based mixed integer linear programming (MILP) network optimization to achieve such a major impact?
DN: At Pioneer, I focused on critical hydraulic-fracturing challenges. WaterWise optimized the allocation of fresh and recycled water across dozens of wellsites by representing the network as a graph. It used a Gurobi-based MILP model to encode supply, demand, and environmental constraints, ensuring every site's needs were met at minimal cost while prioritizing recycled water.
Similarly, SandWise applied a network-optimization framework to the proppant supply chain, factoring in truck capacities, supplier pricing, and travel distances to reduce costs and improve delivery. Together, these tools streamlined resource flow, enhanced operational efficiency, and significantly reduced environmental impact.
AA: You’ve demonstrated proficiency in a range of technical skills, from Bayesian statistics and MLflow to Databricks and convex optimization. Are there any particular tools or skills that you find are most effective for someone working at the intersection of data science and the energy industry today?
DN: Three areas stand out: optimization modeling, probabilistic modeling, and data engineering at scale. Optimization modeling, using tools like Gurobi or Pyomo, is essential for solving complex operational problems in the capital-intensive energy industry. Probabilistic modeling with Bayesian methods, using tools like PyMC, allows engineers to quantify uncertainty for tasks like production forecasting. Finally, data engineering at scale with platforms like Databricks is crucial for integrating vast, high-frequency datasets into real-time analytics.
Mastering these capabilities allows professionals to design scalable solutions that are technically sound and operationally useful.
AA: You've also written several publications, including one on "Embedding Existing Pipelines in Design of CO2 Transportation Networks for Optimal Sequestration Economics." What role do you see data science and AI playing in the future of sustainable energy and carbon capture technologies?
DN: Data science and AI will be central to scaling sustainable energy solutions, particularly in carbon capture. AI can help predict the long-term storage integrity of CO2 and model its movement, as well as design resilient transport networks that adapt to changing supply and demand.
The integration of physics-based models with ML, such as through physics-informed neural networks, will enable faster and more accurate simulations. These innovations will reduce project uncertainty and risk, making large-scale decarbonization both technically feasible and economically viable.
AA: In your current role as a senior data scientist at Intuit, you focus on AI/ML, analytics, and optimization. Do any skills and perspectives you gained from the oil and gas industry translate to the financial technology sector, and vice versa?
DN: Yes, two areas immediately come to mind: optimization and resource allocation, and modeling under uncertainty. As a reservoir engineer, I had to make decisions about the most optimal and cost-effective field development strategies, which mirror the decisions I make at Intuit about which advertising channels are best and how to allocate spending to maximize customer acquisition. The second is the ability to model under uncertainty.
In both fields, you rarely have all the information needed to make decisions. Whether it's forecasting new wells with uncertain geologic factors or building marketing attribution models without complete data, the key is to make informed assumptions and let the data guide you.
AA: Looking back on your diverse journey from your early days in Nigeria through your work in reservoir engineering, all the way to leading innovative data science projects at top tech companies, what mindset or philosophy has helped you overcome obstacles and keep pushing the boundaries of what’s possible in your career? How can others adopt this mindset to achieve their own breakthroughs?
DN: My guiding principle is “learn, adapt, and scale impact.” Every challenge has been an opportunity to acquire new skills, whether it's learning a new programming language or stepping into a new industry. I focus on creating solutions that are not just innovative but also scalable and capable of driving sustained impact.
Others can embrace this mindset by committing to lifelong learning, seeking out interdisciplinary experiences, and viewing unfamiliar problems as opportunities for growth.
AA: What is one piece of advice you would give to an early-career professional, perhaps a petroleum engineer or a data scientist, who is interested in building a career that bridges these two dynamic fields?
DN: I would advise them to build depth in both domain expertise and computational skills, and to actively look for opportunities to integrate them. Professionals who can navigate both engineering realities and data-driven possibilities are the ones who will shape the future of energy and technology. Whether you are optimizing drilling logistics, designing a CCS network, or generating production forecasts, that combination of skills will set you apart.
Stay curious, embrace interdisciplinary projects, and think beyond traditional boundaries.
