Maj. Brandon Delarosa is a U.S. Army Operations Research and Systems Analyst who grew up in a military family and considers King George, Virginia, home. He graduated from the United States Merchant Marine Academy with a Bachelor of Science in Marine Engineering Systems and received a direct commission into the U.S. Army as a Military Intelligence officer in 2014. In 2024, he transitioned to the Army’s Operations Research and Systems Analysis functional area.
Before commissioning, Maj. Delarosa served as a Merchant Marine engine cadet aboard steam, diesel, and diesel-electric vessels. Working alongside ship engineers to maintain complex propulsion and auxiliary systems gave him a hands-on understanding of how large technical systems operate. The experience taught him to learn unfamiliar systems quickly, recognize how individual components interact, and identify opportunities to improve performance. That foundation continues to shape how he approaches complex Army, Navy, and joint problems.
As commander of Headquarters and Headquarters Company, 264th Combat Sustainment Support Battalion, 3rd Expeditionary Sustainment Command, Maj. Delarosa led Soldiers at Fort Bragg and in Kuwait while supporting dispersed logistics operations across the U.S. Central Command area of responsibility. His company supported Immediate Response Force requirements and helped synchronize sustainment for joint and coalition forces.
Maj. Delarosa also served as U.S. Central Command’s Military Linguist Manager. When contractor departures created an urgent language-support gap during combat operations in Iraq, he identified qualified service members across the joint force, both in theater and in the continental United States, to support coalition and joint formations.
As a data scientist with U.S. Army Network Enterprise Technology Command at Fort Huachuca, Arizona, Maj. Delarosa contributed to an unclassified prototype analytics dashboard incorporating retrieval-augmented generation. The dashboard organized global network-status reporting and provided senior leaders with a clearer way to identify trends and emerging issues across the Army’s network enterprise.
Maj. Delarosa is currently pursuing a Master of Science in Operations Research at the Naval Postgraduate School and expects to graduate in June 2027. His research focuses on optimizing the employment of unmanned and attritable systems for distributed military logistics. He combines mixed-integer optimization, geospatial modeling, and reproducible simulation to examine mission assignment, staging locations, throughput, and fleet composition under constraints such as distance, restricted routes, uncertain demand, payload, availability, and cost. His work is intended to give planners a transparent way to compare sustainment options, identify the causes of unmet demand, and understand operational tradeoffs.
Maj. Delarosa is married to the former Rosemarie V. Godoy, and they have a three-year-old son named Andre. Although his next assignment has not been determined, Maj. Delarosa hopes to serve in an Army analytical organization or at the Pentagon, applying operations research and data science to problems facing military decision-makers.
The simulator moves autonomous-resupply planning from qualitative argument to quantitative comparison. A planner builds a scenario that includes staging bases, afloat platforms, and demand points; sets fleet characteristics such as range, payload, and availability; and then watches coverage, delivery throughput, and unmet demand evolve over time. Under the hood, a location-based optimizer assigns each demand point to a logistics node according to explicit prioritization rules. The environment remains grounded in reality: the tool replays actual ship traffic using embedded Automatic Identification System (AIS) data and accounts for weather, restricted airspace, and territorial buffers.
The decision-making capability I care most about is disciplined comparison. Every run exports an audit package including the scenario, assumptions, data version, probability distributions, and results for comparison. The audit package enables planners to compare two courses of action under identical conditions and explain and defend the differences before a decision-maker. The tool is deliberately comparative rather than prescriptive: it does not hand planners an answer; instead, it exposes the trade space among coverage, cost, and redundancy. Random-sampling runs place uncertainty bands around each option rather than presenting a falsely precise single estimate. The embedded AI assistant helps planners build and interrogate scenarios, but every AI-assisted action is approval-gated and logged. In this way, the assistant enhances decision support rather than replacing human decision-making. The next research frontier is applying the tool to contested logistics, where we can model delivery networks and assess how well they hold up when the adversary gets a vote.
Three elements converge at NPS that I believe rarely come together anywhere else. First, the Operations Research curriculum provides a rigorous analytical toolkit, immersing students in optimization, simulation, and statistics at a depth that enables them to formulate and solve complex problems from first principles. The assignment, facility-location, and fleet-mix formulations in my simulator came directly from that coursework. Second, NPS faculty have helped shape the literature the Nation relies on. Their ability to connect mathematical theory to operational problems directly informed my thesis literature review and deepened my understanding of the problem. Third, the students are operators. From what I observed at Converge, every table included people responsible for solving real problems in the force, and students bring their own operational challenges to NPS as I did with distributed sustainment.
Add to that time, compute, and venues like Converge that put a student prototype in front of people empowered to act on it, and you get research that is operationally relevant from the first line of code, instead of trying to reverse engineer at the end.
It compressed what would normally have taken months of cold outreach into just a couple of days. Pitching at Converge meant defending my work in rapid succession before operators, engineers, program managers, and executives, with each group stress-testing a different aspect of it. Operators asked about workflow and trust; engineers asked about data and validation; and transition-focused leaders asked what it would take to move from prototype to program.
The NVIDIA engagement stands out. I had the rare opportunity to brief NVIDIA CEO Jensen Huang on the effort and hear his perspective as someone closely connected to where the technology is heading, along with his feedback on where the project could go next. Beyond that, follow-on conversations with autonomy and defense-technology companies, aerospace primes, and cloud and AI infrastructure teams have focused on exactly what the project needs next: a path toward an operational pilot.
For students, the value lies in access and the speed of feedback. Communicating analysis to decision-makers is a core skill for an Operations Research analyst, and Converge provides a live-fire exercise in doing exactly that. In a single afternoon, I learned which parts of my pitch carried weight and which assumptions did not survive contact.
Three kinds of support stand out, roughly in order of impact. First, an operational pilot: a command, program office, or exercise willing to apply the simulator to a real planning problem. Several partnerships emerging from Converge could make that possible, creating opportunities I would not otherwise have had. Second, data and validation partners with range, payload, reliability, and demand data from actual operations. That information is essential to moving the model from notional to validated. Much of it resides within industry, and the partnerships now taking shape will help us build a richer, more credible model. Third, transition guidance and support: establishing a pathway toward accreditation, developing hardened variants for use at higher classification levels, and creating a fleet-ready user experience that closes the gap between a student prototype and a fielded planning aid.
On the academic side, continued faculty collaboration and support on optimization modeling where down the line we can turn this tool’s contested-logistics ambitions into rigorous thesis results.
I believe the innovation loop gets shorter and the requirements become more realistic. The traditional cycle moves from an operational problem to requirements documents, then to an industry solution, and finally to operational reality, where the original requirements often change. When the problem owner, researcher, and technologist are in the same room, assumptions can be corrected in minutes. Students can serve as translators, fluent in both operations and analytics. They can explain to an engineer why a locally hosted large language model matters and tell an operator what a solver can realistically guarantee.
You can see the effect in the resulting design decisions. I learned where industry is headed and where we might integrate emerging solutions. Industry representatives gain ground truth early from operators recently returned from the field. Researchers learn which constraints are truly binding in operational scenarios, while operators see what is technically possible now rather than what might be possible after another budget cycle.
The cycle also compounds over time. The network formed at Converge does not dissolve when the event ends. Students return to the force carrying those relationships, and many later return to Converge to help guide conversations for the next group. I personally met several NPS graduates working across the Department of Navy and Marine Corps who were able to contribute insights based on what they had previously experienced at Converge.
.jpeg)
