---
title: 'Member of Technical Staff - Low Level & Kernels Capabilities at Preference Model'
canonical: 'https://feeny.ai/job/member-of-technical-staff-low-level-kernels-capabilities-preference-model-san-4r3m43z2fz74'
type: 'job'
last_seen: '2026-09-08'
---

# Member of Technical Staff - Low Level & Kernels Capabilities at Preference Model

- **Company:** Preference Model
- **Location:** San Francisco, CA
- **Compensation:** $200k–$350k
- **Employment:** full-time
- **Work type:** onsite
- **Posted:** 2026-08-25
- **Last confirmed live:** 2026-09-08
- **Apply:** https://jobs.ashbyhq.com/preference-model/85414c2b-d0e1-4c49-a9cb-a2270f214e5a

## Job description

## ABOUT US

Preference Model is building automated ML research engineering.

Existing frontier models are brittle when applied to real-world ML tasks. The present bottleneck is the lack of high-quality RL training environments. Our first step is to build RL environments that reflect real-world complexity, with diverse tasks and robust reward functions.

Our founding team has previous experience on Anthropic’s data team building data infrastructure, and datasets behind Claude. We are partnering with leading AI labs to push AI closer to achieving its transformative potential.

## ABOUT THE ROLE

We’re hiring experienced Machine Learning Engineers for our Low Level / Kernels Capabilities team. The Kernels team builds reinforcement learning (RL) environments at the lowest layers of the stack. Think GPU and accelerator kernels, vector ISAs, codec and crypto primitives, FPGA work, and more. These are the domains where frontier models are weakest, niche paradigms, hardware underrepresented in training data, and open benchmarks that show models lagging.

This role blends research and engineering. It will require you to both develop novel approaches and realize them in code. You will own environments end-to-end: choose the domain, design the tasks, build the scoring and infrastructure, and harden it against reward hacking. Because the tasks run so low in the stack, robust scoring and sandboxing are a real part of the job, making sure a model can't game the timer instead of writing the kernel.

## WHAT YOU WILL DO:

- Design and build low level / kernel-focused reinforcement learning (RL) environments that target a specified model and difficulty distribution.
- Choose which environments are worth building. A strong kernel environment hits several marks:
- Targets a niche or genuinely hard domain;
- Exercises real hardware features (tiling, streaming, async copy, vector ISAs);
- Interesting hardware or simulators (FPGAs, novel accelerators, gem5);
- Research-motivated, grounded in benchmarks where models lag;
- Has a recognized reference to measure against (cuBLAS/FFTW/OpenSSL/etc.);
- Scales into many diverse tasks from a single design.
- Build correctness and performance scoring that's deterministic and can't be gamed: the objective is clear, and the only way to hit it is to actually write the kernel.

WHAT WE ARE LOOKING FOR (QUALIFICATIONS):

- Strong low-level/systems engineering: fluent in C / C++ / CUDA (or an equivalent kernel language), comfortable dropping to assembly when it matters.
- Strong, engineering-quality Python across your prior work, writing production code, automation and deployment scripts, data analysis and plotting (not notebook-only).
- Hardware-aware coding: you write with the silicon in mind, considering memory hierarchy, occupancy, data movement, parallelism, latency vs throughput etc.
- Kernel development experience: you write kernels and optimize them iteratively against a profiler.
- An adversarial mindset: you turn fuzzy goals into robust, ungameable scoring, and you ask "how would a model cheat this?"
- Hands-on work with LLMs
- Ownership and autonomy: you build, debug, and ship end-to-end with minimal supervision.

YOU MAY BE A GOOD FIT IF YOU ALSO:

- Have shipped a kernel that approached SOTA and can explain the remaining gap.
- Have depth in a niche hardware target or ISA: FPGA/HLS, RISC-V Vector, DSPs, SIMD/AVX, TPUs.
- Have depth in an adjacent discipline; HPC/heterogeneous clusters, hardware design (RTL/HDL, HLS), compilers and kernel toolchains (MLIR/LLVM, Mojo, Triton, gem5), or formal verification (Lean, Coq, SMT).
- Read performance and architecture papers and turn them into running code.
- Have open-source contributions others rely on.
- Have a strong competitive-programming background (ideally in a low-level language).
- Have built RL environments, agent harnesses, or evaluation infrastructure.

## WHAT WE OFFER:

- Competitive cash and equity compensation (>90th percentile)
- Ownership and autonomy in a fast moving startup environment
- Opportunity to work with top machine learning engineers
- Health, vision, dental, benefits
- 401K match
- Lunch provided everyday onsite
- Weekly snack orders
- Visa sponsorship & relocation support available

We value diverse perspectives and experiences. If you're excited about this role but don't check every box, we still encourage you to apply.

## About Preference Model

## Company Overview
- **One-liner**: Preference Model builds reinforcement learning (RL) environments that automate ML research and engineering, aiming to teach foundation models how to perform real-world ML tasks.
- **Entity Type**: Private (startup, pre-seed / seed stage – no funding publicly disclosed)
- **Headquarters**: United States (remote-friendly offices also in Canada)
- **Founded**: 2025
- **Founders**: Ning Cao and Jennifer Zhou

## Core Business
- **Primary industry**: Artificial Intelligence / Machine Learning Infrastructure
- **Target customers**: B2B – frontier AI labs and research teams building next-generation LLMs
- **Mission/purpose**: To teach models to perform ML research by creating high-quality RL training environments with diverse tasks and robust reward functions.

## Products & Services
- **Automated ML Research Engineering Platform**: A suite of RL training environments and tools designed to replace the current “brittle” application of frontier models on real-world ML tasks. The platform focuses on reward function design and task diversity to accelerate AI research.

## Market Standing
- **Valuation / Market Cap**: Not disclosed
- **Key Metric**: Total funding not publicly available; headcount is 1 (as of mid-2025) with monthly growth of +66.7%
- **Notable Investors / Partners**: The founding team previously built data and infrastructure at Anthropic, Stripe, and Datology; currently partnering with frontier AI labs
- **Growth Signals**: Website traffic grew +306.9% month-over-month (12,675 monthly visits); headcount growth of +66.7%; strong hiring signals on career page

## Competitive Advantages
- **Founding team pedigree**: Experience from Anthropic, Stripe, and Datology gives deep domain knowledge in data infrastructure and AI safety
- **Focus on a critical bottleneck**: RL environment quality is widely considered the main barrier to advancing frontier models – Preference Model directly addresses this
- **Small, focused team**: Enables rapid iteration and tight collaboration with partner labs

## Strategic Focus
- Automate the entire ML research engineering pipeline, moving from “how” to “what” questions in AI capability development
- Build robust, real-world RL environments that can scale across diverse domains
- Deepen partnerships with leading AI labs to shape the next generation of LLMs

## Why Work Here
- **Culture & impact**: “We are a small team committed to making big impact” – early-stage startup where every hire shapes the product
- **Work environment**: On-site workspace with offices in the United States and Canada (no remote/hybrid details disclosed, but likely flexible for the right talent)
- **Notable perks**: Not publicly listed, but as an early-stage AI infrastructure startup, employees can expect equity, direct influence on technical direction, and close collaboration with frontier AI researchers

## Sources
1. [preferencemodel.com](https://www.preferencemodel.com/) – Company overview, team, mission
2. [linkedin.com](https://www.linkedin.com/company/preferencemodel) – Company details, headcount, growth
3. [builtin.com](https://builtin.com/company/preference-model) – Year founded, office locations, industry tags
4. [github.com/preferencemodel](https://github.com/preferencemodel) – GitHub presence, creation date
5. [jobs.ashbyhq.com](https://jobs.ashbyhq.com/Preference-Model) – Active careers page

## Other roles at Preference Model

- [Member of Technical Staff - ML Infrastructure Engineer, Post-training](https://feeny.ai/job/member-of-technical-staff-ml-infrastructure-engineer-post-training-preference-re11f9e30efq) — San Francisco, CA
- [Member of Technical Staff - Research & Post-training](https://feeny.ai/job/member-of-technical-staff-research-post-training-preference-model-san-francisco-gtttnxrev7hv) — San Francisco, CA
- [Member of Technical Staff - Machine Learning Capabilities](https://feeny.ai/job/member-of-technical-staff-machine-learning-capabilities-preference-model-san-ehsnrydyrvrs) — San Francisco, CA
- [Chief of Staff](https://feeny.ai/job/chief-of-staff-preference-model-san-francisco-evw9xnhwphed) — San Francisco, CA
- [Member of Technical Staff - Machine Learning Capabilities, New Graduates](https://feeny.ai/job/member-of-technical-staff-machine-learning-capabilities-new-graduates-a3tbxd6kyjvy) — San Francisco, CA