Senior Systems Software Engineer - Code Representation & Compiler Infrastructure

Posted: 06/08/2025

Closing Date: 01/10/2025

Job Type: Permanent - Full Time

Job Category:
Information Technology

IR Labs is the innovation lab inside Integrated Research where small, cross‑functional squads chase outsized, industry‑defining opportunities. We operate like a funded startup — rapid sprints, bold experimentation, zero bureaucracy — backed by the global footprint and resources of a public company. Our charter is simple: turn cutting‑edge AI research into products that customers can’t imagine working without. We target the hardest problems in software and then move fast to ship solutions that create 10x impact.

If you thrive on autonomy, crave world‑class technical challenges, and want to see your ideas hit production quickly, IR Labs is your launch pad. Join us and help build the future—one breakthrough at a time.

Do you dream in SSA form and benchmark your work in nanoseconds? At IR Labs you’ll be the founding Systems Software Engineer for Code Representation & Compiler Infrastructure, owning the machinery that turns millions of lines of source into graph‑ready data the rest of our AI stack can reason over. You’ll design custom intermediate formats, build Rust‑powered parsing and diff engines, and squeeze every last cycle so our ML teams can train and serve models in real time.

Because we run like a seed‑stage startup inside a public company, you’ll have the autonomy to rewrite hot paths in SIMD one week and roll out secure, reproducible builds the next—while still tapping enterprise‑scale resources and customers. If compilers, low‑latency data pipes, and zero‑bureaucracy shipping sound like your idea of fun, come join us and help redefine how software understands itself.

This is a remote role, but candidates located in Denver, CO are strongly preferred.

What You’ll Do

• Architect and implement the core code-representation engine in Rust, transforming monolithic repositories into rich intermediate forms ready for graph analytics and ML—without compromising millisecond-level latency.
• Design custom IRs and optimization passes that capture syntax, control/data flow, and semantic metadata; surface incremental diff graphs that track every code change over time.
• Build ultra-high-throughput parsing and analysis pipelines using lock-free concurrency, SIMD/vectorization, and optional GPU offload to compress multi-hour workloads into seconds.
• Expose clean, versioned APIs and libraries (gRPC / Flat Buffers / Arrow) that allow LLM and Graph MLEs to consume features, embeddings, and provenance data effortlessly.
• Continuously profile, benchmark, and harden the pipeline—leveraging perf, eBPF, sanitizers, and fuzzers—to guarantee security, determinism, and predictable memory footprints.
• Collaborate closely with ML engineers to iterate on feature schemas, optimize data-access patterns, and co-design benchmarks that reflect real-world model performance needs.
• Upstream improvements to open-source compiler or analysis tool chains where strategic, and champion best-practice Rust patterns across the engineering org.
• Automate CI/CD and release workflows with reproducible builds, container images, and signed artifacts, ensuring rapid yet trustworthy iteration.

Qualifications

• 8 + years building high-performance systems software; 4 + years hands-on Rust in production, including contributions to compiler/runtime or deep systems libraries.
• Mastery of compiler fundamentals—lexing/parsing, IR generation, SSA, alias analysis, code-gen—and practical experience extending LLVM, MLIR, Cranelift, or similar.
• Track record processing codebases >10 M LOC with incremental or distributed compilation/analysis strategies.
• Demonstrated ability to squeeze performance via SIMD, cache-aware memory layout, NUMA tuning, and (optionally) GPU or SPIR-V compute.
• Fluency with graph-structured program representations (AST/CFG/PDG) and diff/patch algorithms; comfort persisting and querying large property graphs.
• Deep tooling chops: perf, Flamegraph, Valgrind, eBPF, sanitizers, and AFL/lib Fuzzer; proven incident-free track record under strict performance budgets.
• Familiarity with ML data pipelines—vector stores,…