About Normal Computing

Normal Computing builds silicon that turns thermal noise from an obstacle into a computational resource. Conventional chips spend most of their energy forcing determinism onto physics; ours compute with it. Stochastic, in-memory, asynchronous: the result is 10-100× more AI inference per dollar, per watt.

We co-design the full stack: AI-native EDA systems in production with the world's largest semiconductor companies, and the advanced ASICs they make possible. Backed by $85M+ from the world's leading deep-tech investors and built by scientists, engineers, and operators from the labs that built modern computing.

Normal works as one team across New York, Silicon Valley, London, Copenhagen, and Seoul. We hire people who want the hardest version of their craft, across every discipline, at every seniority.

The Role

Our EDA tool accelerates the design and verification of silicon. It integrates with the engineer's workflow to assist with design, verification, and debugging, and uses AI to generate stimulus, tests, SystemVerilog assertions, and other verification artifacts.

The hard part of that is data, and most of the data we need is not in a format that is easy to train on. The verification artifacts that would teach our agents are locked inside customer environments, paywalled behind standards bodies, or simply never written down. So this role is not primarily about finding data. It's about manufacturing it: generating synthetic training data with programmatic ground truth, mining our own agents' runs for high-quality trajectories, and negotiating access to the real customer data that nothing else can replace. You will own that pipeline end to end and partner directly with the ML/post-training and eval teams, because the only definition of success here is moving a number on our eval harness.

You'll work alongside verification engineers who set the standard for "good," ML engineers who help tune the generation loop, and pipeline engineers who keep the data organized and versioned. The strategy for what data we build, mine, and acquire is yours.

What You'll Own

• Model Improvement: Your main responsibility is making our models better at hardware design, verification, and EDA workflows by any means possible.

• The Data Flywheel: Own the data flywheel from our own agent runs: rejection sampling, distillation, and mining eval-passing trajectories so each model round produces the training data for the next.

• Data Acquisition: Identify, evaluate, and acquire datasets relevant to hardware design, verification, and EDA workflows, with a focus on data that drives measurable improvement in AI agent performance. Assess sources for quality, coverage, licensing, and compliance before ingestion.

• Quality Standards: Partner with verification engineers to define rubrics, curate golden reference examples, and tell when the pipeline is producing convincing-looking garbage.

• Pipelines & Lineage: Operate data ingestion pipelines, monitor for quality regressions and coverage gaps, and maintain a structured catalog of data sources, acquisition strategies, and lineage.

• Customer-Data Partnerships: Negotiate access on customer infrastructure (on-prem and federated), handle redaction and IP constraints, and where direct access isn't possible, build external replicas of a customer's environment that preserve the structure of their specs and testbenches without exposing their IP.

• Team Building: As the Data team scales, manage engineers across synthetic data, verification SME curation, data infrastructure, and forward-deployed data engineering.

What Makes You a Great Fit

• You've built or used a data flywheel: model outputs, curated, into the next training round

• You approach data acquisition as an engineering problem: systematic, measurable, and outcome-driven

• You've shipped a synthetic-data or training-data pipeline that produced a measurable downstream model improvement you can describe by number, not vibes

• You can evaluate data quality independently, spotting noise, bias, and gaps without needing someone to tell you what to look for

• You're comfortable working across multiple technical roles and synthesizing feedback from domain experts, ML engineers, and pipeline engineers

• You're organized and documentation-minded: you track provenance, ownership, and lineage as a matter of habit

Bonus Points

• Experience acquiring data from a variety of sources, both paid and unpaid, and managing vendor relationships

• Familiarity with SystemVerilog, Verilog, and UVM

• Background in code-model or agent training-data pipelines (e.g. SWE-bench-style data, code-model post-training)

• Experience with automated data collection, web scraping, or corpus curation at scale

• Prior work in a startup or fast-moving research environment where the data strategy was still being defined

Equal Employment Opportunity Statement

Normal Computing is an Equal Opportunity Employer. We celebrate diversity and are committed to creating an inclusive environment for all employees. All qualified applicants will receive consideration for employment without regard to race, color, religion, sex, sexual orientation, gender identity, national origin, disability, veteran status, or any other legally protected status.

Accessibility Accommodations

Normal Computing is committed to providing reasonable accommodations to individuals with disabilities. If you need assistance or an accommodation due to a disability, please let us know at accommodations@normalcomputing.com.

Privacy Notice

By submitting your application, you agree that Normal Computing may collect, use, and store your personal information for employment-related purposes in accordance with our Privacy Policy.