Rust Engineer

Hello,

 Role - Rust Engineer

Duration - 6+ Months

Location - Seattle, WA

Job Summary

As a member of the Region Flexibility Migration (RFM) team, you will help drive Amazon''s DRAM memory optimization initiative: reducing the memory footprint of Amazon''s service . Rustification is one of the primary levers: rebuilding memory-bound and latency-critical services  as memory-safe Rust services to eliminate JVM runtime overhead ( heap/metaspace, JIT), unlocking proven gains of 50%+ memory reduction, 40%+ latency improvement, and 2-4x throughput.

You are expected to have deep, end-to-end knowledge of how a Rust service is built, validated, deployed, and optimized: enough to drive results through the agents, exercise the engineering judgment they cannot (architecture, type hierarchies, equivalence sign-off), diagnose failures when the agents get stuck, and contribute improvements back to the framework''s skills, steering, and CI enforcement. The agents accelerate the work; your expertise directs it.

This is a high-autonomy role. You will own a migration end-to-end:  through implementation, equivalence validation, performance/memory optimization, and safe cutover, while coordinating with upstream and downstream teams and codifying reusable learnings that let the DRAM program scale.

Key Responsibilities

Drive Java-to-Rust migrations as a DRAM-reduction lever: apply the decision framework to identify genuinely memory/latency-bound services (JVM overhead as a structural ceiling), rank APIs by traffic, and own each migration end-to-end from analysis through cutover.

Bring deep system expertise to the agent-driven workflow: define type hierarchies, module boundaries, and API contracts; drive and unblock the Rustify agents rather than merely operating them; and contribute improvements back to the framework when you hit friction.

Establish oracle-based correctness and prove functional equivalence: capture production request/response pairs as the source of truth and validate the Rust implementation via progressive zero-diff testing (RustifyComparator) from small samples to full 1000+ scale, distinguishing real bugs from acceptable non-determinism.

Implement on the target stack and optimize for memory: build on Smithy Rust Server SDK, ECS Fargate (ARM64/Graviton), CargoBrazil, CDK with @amzn/pipelines, and CloudAuth; then profile hotspots, stress test, and quantify DRAM/latency savings with no functional regression.

Execute safe rollout and operations: shadow deploy in Gamma and Prod to prove equivalence on live traffic, drive gradual cutover (canary to 100%) with rollback at each stage, stand up observability (dashboards, alarms, runbooks), and align API compatibility with upstream/downstream teams.

Codify reusable learnings: migration guides, runbooks, and friction logs

A Day in the Life

Your morning might begin by running analysis against a Java service to confirm its memory/latency profile justifies a rewrite, ranking its APIs by traffic, and authoring the migration plan that the framework uses to derive its work units. You review it, define the module boundaries and type hierarchies yourself, and kick off the first rebuilds.

By mid-day, a unit is stuck after the agent''s self-correcting loop exhausts its attempts on a serialization edge case. You inspect the behavioral spec and the oracle diff, fix the logic yourself, and re-run zero-diff validation until it passes at full scale.

After lunch, you review a Gamma shadow-deploy report: 99.8% match, remaining diffs traced to non-deterministic ordering you document as acceptable. You present the DRAM/latency results and phased cutover plan to an upstream team, aligning on rollout timing and rollback triggers. Before wrapping up, you capture the day''s friction into a log and open a CR to improve the framework so the next migration hits less resistance.

You operate with high autonomy, own your migrations end-to-end, and thrive on making services leaner and cheaper to run.

About the Team

The Region Flexibility Migration (RFM) team is part of Amazon''s Intelligent Cloud Hosting (ICON) organization. Formed in 2024 to proactively discover and mitigate issues teams face when migrating services across AWS regions, RFM has built deep expertise in AWS service architectures, infrastructure optimization, and large-scale migration programs. In 2026, RFM expanded its mission to a critical DRAM memory optimization initiative targeting significant memory reduction across Amazon''s service fleet, with rustification as a key lever. We are looking for an exceptional engineer with strong Rust and memory-optimization expertise to drive this work at scale.

Basic Qualifications

Bachelor''s degree in Computer Science, Engineering, or equivalent technical field.

5-7+ years building and operating backend services in large-scale production environments, with demonstrable end-to-end ownership across design, implementation, testing, deployment, and operations.

End-to-end experience building, deploying, and maintaining production Rust applications: solid command of ownership/borrowing, lifetimes, traits, error handling, and an async runtime (e.g., Tokio); able to write, review, and debug Rust, not just read it.

Hands-on performance and memory optimization experience in production services: profiling, benchmarking, and reducing memory footprint, with a working understanding of how memory optimization is achieved (e.g., allocation patterns, data layout, eliminating runtime overhead).

Demonstrated experience migrating or rewriting services across languages or runtimes while preserving API/wire-format compatibility and proving functional equivalence.

Familiarity with the Rust service target stack or close equivalents: Smithy Rust Server SDK, ECS Fargate (ARM64/Graviton), CargoBrazil, and CDK with @amzn/pipelines.

Proficiency working hands-on with AI coding agents as a primary development workflow: driving multi-step/multi-agent workflows, structuring work within human-defined boundaries, reviewing agent output critically, and knowing when to intervene versus delegate.

Strong Java/JVM experience (or equivalent in another source language), including the runtime cost model ( heap/metaspace, JIT), sufficient to judge migration candidates and faithfully re-express existing service logic and framework behavior (e.g., Coral, Spring, Guice).

Preferred Qualifications

Measurable DRAM/latency/throughput improvements achieved in production distributed systems through rewrites, memory optimization, or runtime changes.

Experience with oracle/differential testing and shadow deployment (traffic mirroring, response diffing, statistical equivalence) as a validation and cutover strategy.

Experience contributing to AI agent frameworks or automated engineering workflows: not just using them, including skills, steering, or CI enforcement to prevent convention drift.

Experience with Rust interop/FFI (JNI/bindgen) and incremental, per-API migration where Rust and Java services coexist behind stable boundaries.

Experience with Amazon internal validation and operational tooling (profilers, stress-testing frameworks, CloudWatch, X-Ray) and standing up observability for a new runtime.

Strong written and verbal communication: able to present equivalence and performance results and frame proposals around fleet cost / IMR / DRAM reduction to service teams and leadership.

Thanks and Regards,

Shikha Sharma

Desk: (551) (900) (7008)