Project: AMIC, autonomous vehicle data · Budget: $25M over 5 years · Engagement: Apr 2022 to present

  • 1,200+ TB of autonomous vehicle data migrated to the new FedRAMP-compliant system
  • 3 people delivering what should have required eight in a matrix org
  • Only 1 contractor extended through transition, out of the entire program team

Why this environment is different

Lincoln Laboratory isn’t a typical government IT environment, and that’s exactly what made this assignment difficult. It’s the DoD’s largest federally funded R&D center: 4,500 staff, a $1B+ annual budget, 90% DoD-funded, operating continuously since 1951. Its mandate is to solve national security problems the government’s existing resources cannot.

It also runs as a matrix organization. Researchers contribute across multiple programs at once, often at 10 to 25% allocation. Each program is built from scratch with bespoke tooling and no shared cloud infrastructure. When staff rotate to higher-priority programs, which is routine here, institutional knowledge leaves with them. The Army customer, GVSC, expected consistent delivery regardless.

AMIC (Autonomous Mobility through Intelligent Collaboration) is the Army’s purpose-built data infrastructure for autonomous vehicle field data. The objective: establish U.S. operational dominance in military autonomous vehicles. Reaching “program of record” status, annual Army maintenance funding, means the software outlives its original contract. That outcome is rare at Lincoln Laboratory.

The problem

AMIC had been funded since September 2020. When we engaged in April 2022, the platform already had production users on software that couldn’t be reliably built, maintained, or extended. The underlying technology was a decade out of date, dependency chains were severe, and the build process was broken, which trapped every incoming contributor in onboarding instead of delivery.

I came into a broken legacy system built from technology and patterns that were 10 years outdated, with extremely long dependency chains. The build process was broken, and the learning curve to get up to speed was extremely high.

Two to three months in, the original Principal Investigator left. As a contractor, we trained the incoming PI, a full-time Lincoln Lab employee, while the team kept rotating to other programs. With no lab-wide cloud strategy or internal AWS expertise at the time, every architectural decision had to be built from first principles inside a FedRAMP-compliant, CUI-handling government cloud, all while keeping existing production users running.

The approach

The hard call was to stop patching the old system and build the right one in parallel.

A hard-fork decision

We assessed the legacy system and determined that continued maintenance couldn’t meet security requirements or the program’s operational demands. We built a replacement in parallel while keeping the existing system running for production users throughout.

Ground-up architecture on AWS GovCloud

A resilient EKS cluster with a Kubernetes backend, CI/CD pipelines, and test-driven development. Built for a small team to operate and iterate quickly, not dependent on the staffing level the original program assumed.

FedRAMP and CUI compliance at scale

FedRAMP authorization, Controlled Unclassified Information handling, and Army-grade security posture, without sacrificing throughput across a 1,200+ terabyte dataset.

A lab-wide AWS strategy, the multiplier

We worked through the lab’s administrative complexity to establish cloud strategies and security postures that now benefit programs across the entire institution. Other programs no longer build cloud infrastructure from scratch, a gap that had cost them time and budget for years.

Knowledge transfer that actually transferred

We trained multiple rotating PIs as they cycled through the matrix org, onboarded new developers fast enough to contribute without a long ramp, and are now training the Army-side transition team, the final step before the program-of-record handoff.

What we delivered

Six workstreams. Three people. One environment where most contractors don’t finish.

  • Cloud architecture (AWS GovCloud). A resilient EKS cluster with CI/CD and TDD, production-grade infrastructure delivered by three people.
  • FedRAMP & CUI compliance. Full government cloud security posture, every Army requirement met at terabyte scale.
  • 1,200+ TB migration. Multipart upload and download infrastructure for large-file handling in a compliant environment. Full dataset transferred without production disruption.
  • Cost modeling & burn-down. Budget burn-down models, the metric Army customers use to evaluate contractor performance at every milestone. Discipline maintained across multiple extensions.
  • Lab-wide AWS infrastructure. Reusable cloud strategies and security postures now available across Lincoln Laboratory.
  • Dual-track operations. Legacy incident response for live users in parallel with the full rebuild. No production outages during a complete system replacement.

The outcome

Successful program transitions at Lincoln Laboratory are the exception. The matrix organization, rotating personnel, and institutional complexity mean most programs end a contract requiring the receiving team to reconstruct significant context. AMIC will reach a full, successful handoff to the Army, the outcome GVSC requires to grant program-of-record status.

When the initial contract expired, one contractor was extended to carry the program through transition: the team that held full system context, had trained every PI rotation, built the infrastructure to be maintainable, and owned the direct Army customer relationship. That was us.

This is a program that will reach a successful handoff, which is not something that happens all the time within the Lincoln Laboratory.

When the 2025 DOGE federal funding cuts eliminated other programs, AMIC stayed on the books for completion. A direct signal of its strategic importance to the Army’s autonomous vehicle roadmap, and of a program built to demonstrate real operational value rather than just spend a budget.

The results

  • $25M Army program rescued, rebuilt, and carried to a successful transition
  • 1,200+ TB migrated to new FedRAMP-compliant architecture
  • Only 1 contractor extended through transition
  • $1B+ org now benefits from lab-wide AWS strategies
  • Survived DOGE kept funding when other programs were cut
  • 3 people did what should have required eight

AMIC isn’t a delivery story. It’s a proof of operational maturity under conditions designed to produce failure: rotating personnel, institutional complexity, a broken legacy system, and federal funding pressure. It reached a successful handoff because the infrastructure was built to be owned, not managed. That is what separates operationally mature businesses from fragile ones.