Senior/Staff Software Engineer, Compilers
- Discuss this with your agent
- Open in Claude
- Open in ChatGPT
Prepare application
- Draft this with your agent
- Open in Claude
- Open in ChatGPT
Role details
Tech stack
Job description
Our compiler, built in Python on xDSL, represents designs and manufacturing work as typed intermediate representations (IRs). We call it a synpiler: a synthesizing compiler that combines transformations with synthesis steps to resolve missing details as a design becomes a manufacturing program.
Today, the compiler resolves work to methods and block classes; separate tools generate machine instructions. You’ll bring those stages into shared compiler representations, interfaces and verification, including decisions that depend on measurements taken during production.
You’ll own compiler representations, lowering and verification. FABOS maintains production state, plans and authorizes work, and reconciles physical results. Machine endpoints retain final binding, lowering and local control using the selected block’s context. Compiler verification checks representation validity; physical measurements support acceptance of the device against its requirements.
AI fluency is central: use agents throughout development while owning the decisions, code and results you accept.
What You’ll Do
- Build xDSL dialects, analyses, synthesis steps, transformation passes, verifiers and lowering pipelines.
- Represent physical effects, measurement-dependent control flow and unresolved decisions; preserve source process graphs while refining derived IR.
- Define instruction-generation targets and interfaces with FABOS and machine endpoints, making explicit what must be resolved before dependent work can execute.
- Integrate CAM, toolpath and machine-specific instruction generation through typed interfaces, using the selected block’s configuration, calibration and limits during endpoint lowering.
- Specify and test the properties transformations must preserve, including required ordering and evidence; invalidate affected artifacts when target assumptions change.
- Work with Process, CAM, motion, planning and Quality engineers on requirements, acceptance criteria and fab integration; connect generated instructions to requirements and acceptance evidence.
- Use AI throughout investigation, design, implementation, testing and documentation; inspect changes and verify results.
Requirements
Software Engineering, Artificial Intelligence (AI), Entrepreneurship, Manufacturing Systems, Electromechanics, DSL (Digital Subscriber Line), Python Programming/Scripting Language, Manufacturing, Machine Tool, Production Planning, * Production software ownership, typically 8+ years for the senior target; productive Python and strong Go or C++, with readiness to work in Python/xDSL.
- Implementation depth in compilers: typed IRs, analyses, transformations, effects, control flow, verification and progressive lowering.
- Strong AI fluency: provide context, challenge proposals, inspect code and meaningfully test accepted work.
- Evidence of simplifying, testing, operating and safely evolving systems, with clear interfaces, technical writing and explicit correctness conditions for compiler transformations.
- Ability to turn user needs and expert knowledge into requirements, working with process, geometry, motion and planning specialists.
- Willingness to work directly with fab equipment; hardware experience preferred.
- BS in CS or a related field.
Also Useful
- MLIR or xDSL implementation experience, including dialects and rewrite or conversion infrastructure; CIRCT, async or Transform dialects, or equality saturation.
- Program synthesis, partial programs, constrained search, or specification and verification using SMT solvers, TLA+, Lean, Coq or property-based testing.
- CNC/CAM, G-code, trajectory generation, or integration with C++ motion-control code and gRPC services.
About the company
Atomic Machines is building the Matter Compiler, the first in a new class of AI-native, all-digital manufacturing systems that build working machines directly from code. The Matter Compiler (MC) is aimed at micro-machines: motors and gears the size of a grain of sand, robots small enough to work inside the body, cooling built directly into the chips that run AI, and the many devices like them that have been waiting on a way to build them. Its first product is PrimeSwitch, an electromechanical power relay the size of a button battery that carries 150 amps continuous current and opens in 50 microseconds, about a thousand times faster than a conventional contactor. Founded by Jeff Holden - serial entrepreneur, ex-Amazon SVP and Uber’s first Chief Product Officer - the company has raised $250 million to date.
Atomic Machines is based in the San Francisco Bay Area, with offices in Emeryville and Santa Clara.
Apply for this position
This job is hosted externally. Click below to view the full posting and apply.
Prepare application
- Draft this with your agent
- Open in Claude
- Open in ChatGPT
Good distractions
Loading talks and stories from around this role…