As artificial intelligence advances across the defense and autonomy sectors, SAIC is positioning itself as a leader in operationalizing AI at the tactical edge. Jay Meil, SAIC’s chief data scientist, explains how the company is integrating AI to enhance maritime Intelligence, Surveillance and Reconnaissance (ISR), drive human-machine teaming, and optimize data processing at scale.

BIO: Jay Meil is the chief data scientist for SAIC’s Artificial Intelligence Innovation Factory, where he directs AI strategy and solutions across multiple domains. In 2021, Meil was recognized with an SAIC research fellowship for his work in multi-modal data fusion. Meil has more than 18 years of experience designing and implementing applied mathematical models to drive data-centric decision-making at scale. He is passionate about developing AI solutions that facilitate the transformation of multi-INT data into actionable intelligence in support of a common operating picture across defense and national security agencies

“AI is a buzzword right now, but we focus on automation, machine learning, deep learning and areas like neuro-symbolic AI, generative AI and agentic AI,” Meil said. “We’re working to converge these capabilities to make AI safer, more understandable and ultimately more useful.”

Over the past four years, SAIC has pursued a deliberate strategy to move beyond “sandbox” AI experiments and focus on real-world operationalization. “If you can’t integrate and deploy AI at the tactical edge, it’s just a science project,” Meil noted. “Our AI must achieve three things: reduce cognitive load for operators, increase decision speed and create human-machine teams that act as force multipliers.”

This philosophy is particularly relevant in ISR, where decision-makers are expected to cover vast operational areas with limited personnel and resources. By leveraging AI for pattern recognition, object identification and signal detection, SAIC enables operators to focus on higher-order strategic reasoning while AI handles repetitive and data-intensive tasks.

AI DEPLOYMENT AT THE TACTICAL EDGE 

SAIC’s approach to AI deployment emphasizes integration, explainability and security. Meil describes their model as “best-of-breed,” meaning SAIC selects and customizes the best available AI models from trusted sources and then integrates them within secure operational frameworks.

“We don’t build monolithic AI systems from scratch,” Meil said. “Instead, we start with a solid, validated foundation—whether open-source or proprietary—and adapt it to mission-specific needs. That means embedding security controls, aligning dependencies and optimizing the model for different deployment environments.”

One of SAIC’s breakthroughs is the ability to deploy AI models directly onto maritime platforms. “We can bolt on AI models to existing sensor systems—even those with lower-fidelity cameras—and perform real-time inference at the edge,” Meil explained. This capability allows AI to process and analyze ISR data on-board a vessel, rather than relying on cloud-based computing.

HUMAN-MACHINE TEAMING AND AI EXPLAINABILITY

While AI excels at identifying signals within large datasets, human operators remain the ultimate decision-makers. “We believe in human-machine teaming where AI handles the ‘grunt work’—detecting patterns and anomalies—while humans oversee strategic interpretation,” Meil said.

However, AI adoption within the intelligence and defense sectors faces a key challenge: trust. “The biggest barrier to adoption isn’t technical—it’s human,” Meil emphasized. “If operators don’t understand or trust the AI, they won’t use it. That’s why we prioritize explainability.”

SAIC leverages retrieval-augmented generation (RAG) and knowledge-augmented generation (KAG) to ensure AI-generated insights align with human logic. “We design AI systems that can show operators exactly how they arrived at a conclusion,” Meil said. “For example, if AI flags an ISR target, it doesn’t just provide a confidence score—it walks the operator through its reasoning, step by step.”

BUILDING A SCALABLE DATA INFRASTRUCTURE

Despite AI’s potential, its effectiveness hinges on the availability of high-quality data. Efficacy is directly tied to the quality of the underlying data layer. Meil argues that the biggest obstacle to AI adoption isn’t the algorithms—it’s that data fabric that underlies them.

“To scale AI effectively, we need a foundational data layer capable of ingesting, harmonizing, labeling and curating data from multiple sources,” Meil said. “At SAIC, we’ve focused on this problem through projects like the Joint Fires Network [JFN], where we integrate disparate data streams into a common data layer.” JFN is a U.S. military initiative designed to integrate and share targeting data across multiple domains—air, land, sea, space, and cyber—using a networked, AI-driven approach to improve battlefield coordination and decision-making. AI is critical to JFN as it enables real-time data fusion, automated target recognition and predictive analytics, allowing operators to process vast amounts of sensor data quickly and enhance precision in joint military operations.

The hybrid data fabric approach allows SAIC to fuse structured and unstructured data sources, including ISR feeds, sensor data and signals intelligence, into a coherent operational picture. “We’re moving toward a decentralized model where individual platforms—whether unmanned maritime vessels or aircraft—can operate independently, process data at the edge and later synchronize across a distributed network,” Meil explained.

SCALING AI FOR MARITIME ISR AND BEYOND

SAIC envisions a future where AI-enhanced ISR systems autonomously identify, classify and track objects across multiple domains. This is especially critical in maritime security, where threats often emerge in contested and GNSS-denied environments.

“In ISR, we need AI models capable of reasoning across different intelligence sources,” Meil said. “For example, if we detect a ship, we can correlate multiple data points—its AIS beacon, radar signature, visual imagery and electronic emissions—to build a high-fidelity profile. AI assists in fusing this information together, allowing human operators to make better-informed decisions.”

The company is also pushing AI-driven ISR to new frontiers by developing lightweight models optimized for low-power platforms. “We don’t want to rely on massive cloud infrastructure,” Meil noted. “Instead, we’re focused on making AI models smaller, more efficient and capable of running on constrained platforms such as autonomous maritime drones.”

THE FUTURE OF AI IN AUTONOMY AND ISR

Looking ahead, SAIC is exploring new AI architectures that enhance operational resilience. “We’re moving toward object-based intelligence [OBI], where AI creates persistent, enriched data objects rather than just one-off detections,” Meil explained. “For example, if we track a ship over time, AI continuously refines its classification using multiple sensor inputs. This cumulative learning approach improves accuracy and mission effectiveness.”

Additionally, SAIC is researching federated learning models that allow multiple ISR assets—such as unmanned underwater vehicles (UUVs) or aerial drones—to share AI training data without centralized cloud dependencies. This means AI models can evolve and improve autonomously, even in disconnected environments.

Meil believes these advancements will have far-reaching implications across defense, maritime security and commercial ISR applications. “The missing piece in AI isn’t technology—it’s adoption,” he said. “With the right infrastructure, explainability and human-machine collaboration, we can scale AI to levels previously thought impossible.”