Research Engineer, RL Frontiers

Etched.ai; San Jose (CA); On-site; USD 360,000 - 440,000; Full time

Benefits offered by this job

Medical, dental, and vision packages

Housing subsidy

Relocation support

Wellness benefits

Daily lunch and dinner in our office

Unlimited compute budget

Job summary

Etched.ai is building frontier AI systems in San Jose, aimed at advancing RL-centric development for kernel, ASIC, and production pipelines. You will train frontier models to automate complex hardware and software tasks using proprietary simulators and environments.

You will own the post-training stack end to end, build RL environments with hardware engineers, and craft data mixes and benchmarks to push state-of-the-art performance while ensuring ROI.

Qualifications

• Hands-on post-training experience, including SFT and RL (e.g., PPO, GRPO), and shipped models.
• Taken a model from bad to good on a task it wasn’t built for, ideally with scarce data, as an early post-training hire.
• Strong PyTorch and distributed-training engineer who checks rollouts and numbers before trusting results.
• Understand systems RL well enough to speed up loops, not just improve quality.
• Entrepreneurial and capable of making good calls under compute and data constraints.

Responsibilities

• Own the post-training stack completely end to end.
• Build RL environments with ASIC, kernel, and supercomputing engineers, rewarded by simulation, formal tools, tests, and hardware measurements.
• Create custom data mixes that unlock frontier performance on relevant tasks.
• Create benchmarks that drive real productivity in the models and avoid reward hacking.

Skills

Post-training RL

SFT

PPO/GRPO

PyTorch

Distributed training

Model evaluation

System optimization

Tools

verl

OpenRLHF

vLLM

SGLang

TRL

Job description

Job Summary

Etched is building AI systems that accelerate its own development. As a Research Engineer on the RL Frontiers team, you will train our proprietary frontier models to automate kernel development, ASIC engineering, production pipelines, and 3-D hardware operating systems.

You will use our proprietary simulators and environments to create custom data mixes and post-train models to SOTA performance on our hardest engineering tasks.

Key Responsibilities

• Own the post-training stack completely end to end.

• Build RL environments with ASIC, kernel, and supercomputing engineers, rewarded by simulation, formal tools, tests, and hardware measurements.

• Create custom data mixes that unlock frontier performance on relevant tasks.

• Create benchmarks that drive real productivity in the models and avoid reward hacking

You may be a good fit if you

• Have hands-on post-training experience, including SFT and RL (e.g., PPO, GRPO), and have shipped models people used.

• Have taken a model from bad to good on a task it wasn’t built for, ideally with scarce data, for example as an early post-training hire at an applied AI company.

• Are a strong PyTorch and distributed-training engineer who reads the rollouts and checks the numbers before trusting a result.

• Understand the systems side of RL well enough to make the loop faster and cheaper, not just better.

• Are entrepreneurial and make good calls under compute and data constraints. Former founders are a plus.

Strong candidates may also have experience with

• RL or post-training frameworks (e.g., verl, OpenRLHF, TRL, prime-rl) or inference engines (vLLM, SGLang).

• Agentic, multi-turn RL with verifiable rewards.

• Distillation and training small, specialized models for production.

• ASIC design or verification, accelerator kernels, or large-scale compute infrastructure.

Benefits

• Medical, dental, and vision packages with generous premium coverage

  • $500 per month credit for waiving medical benefits

• Housing subsidy of $2,500 per month for those living within walking distance of the office

• Relocation support for those moving to San Jose (Santana Row)

• Various wellness benefits covering fitness, mental health, and more

• Daily lunch and dinner in our office

• Unlimited compute budget subject to ROI justification

Base Compensation Range

• $400K+

• Plus Significant Equity

How we’re different

Etched believes in the Bitter Lesson. We are the first inference-focused frontier AI system, betting early on transformer and transformer-like architectures and on increasing model sizes. Our addressable market is the entirety of inference, unlike many of our competitors.

We are a fully in-person team in San Jose (Santana Row), and greatly value engineering skills. We do not have boundaries between engineering and research, and we expect all of our technical staff to contribute to both and work across disciplines as needed.

About Etched

Etched is building hardware for frontier intelligence. We co-design chips, racks, software, and manufacturing to deliver best-in-class throughput and latency across both prefill and decode workloads. Our first products are heavily focused on inference. Backed by hundreds of millions from top-tier investors and staffed by leading engineers, Etched is redefining the infrastructure layer for the fastest growing industry in history.