Member of Technical Staff (AI research)

Kindredventures; San Francisco (CA); On-site; USD 180,000 - 240,000; Full time

Benefits offered by this job

Meaningful equity (4-year vest)

Health insurance

401(k) with company match

Meals in office

Job summary

Fearn is seeking an experienced AI researcher to join our team in San Francisco. You will train models, design inference pipelines, and push the frontier of visual-language models for the patent lifecycle.

You will collaborate across research and product teams to deploy production-ready solutions, optimize for latency and cost, and contribute to scalable AI infrastructure with a focus on post-training techniques and GPU optimization.

Qualifications

• Proficient in Python and ML frameworks.
• Post-training techniques expertise (RLVR).
• Quantization, batching, distributed training experience.
• GPU infrastructure and optimization experience.
• Published research in top venues is a plus.
• Ability to bring models from prototype to production with teams.

Responsibilities

• Build and ship core parts of Fearn's platform from training models to production.
• Work across language, image, and multimodal models.
• Collaborate with infra team for latency, throughput, and cost.

Skills

Python

ML frameworks (JAX, PyTorch)

Post-training techniques (RLVR)

Quantization & batching

GPU optimization

Research publications

Job description

Careers

/

Member of Technical Staff (AI research)

Member of Technical Staff (AI research)

You will build and ship core parts of Fearn’s platform from deploying our custom models to building frontend experiences.

Role

Member of Technical Staff (AI research)

Location

San Francisco, California

Job TYPE

Full-Time

About the Role.

We're looking for an experienced AI researcher to join Fearn as we build out our key features. You'll play a crucial role in training models and designing inference pipelines, pushing the frontier of VLMs for the patent lifecycle.

What you'll work on.

• Building and shipping core parts of Fearn's platform from training custom models to post-training to serving them at production scale. You'll work across language, image, and multimodal models.

• Pushing the frontier of what our models can do, using post-training techniques like RLVR, and bringing new models from prototype to production.

• Staying close enough to infra to get your own work into production. You'll collaborate with the AI infra team on quantization, batching, multi-model serving, and optimizing for latency, throughput, and cost.

What you'll bring.

• Expertise in Python and ML frameworks (e.g. JAX, PyTorch).

• Expertise in post-training techniques (e.g. RLVR).

• Experience with quantization, batching strategies, and distributed training.

• Experience with GPU infrastructure and optimization.

• Published research (top venues like NeurIPS or CVPR are a plus, but not required).

• Experience with collaborative problem solving where you can bring models from prototype to production alongside research and product teammates.

Bonus.

• Production AI serving with vLLM, SGLang, or similar.

• Kernel-level optimization.

• Monitoring, logging, and observability tooling for AI inference.

• Strong GCP experience, particularly Compute Engine, GKE, Vertex AI, or Cloud Run.

What we offer.

• Competitive salary and meaningful equity, calibrated to experience (4-year vest).

• Health insurance, 401(k) with company match.

• Meals in office.

About our interview process.

We run an efficient and thorough interview process to ensure a good fit on both sides. That being said, we will try our best to make the process as fast as you need it to be. Please see details below:

Member of technical staff.

Introductory chat: We will tell you a bit about Fearn, learn about you and your prior experiences, and answer any questions you might have. The goal is to assess whether or not we are aligned.

Take home system design interview with follow up: You will have a few days to work on a couple of system design problems. Afterwards, we will meet to discuss your implementation to understand your thought process. Our goal is to understand how you reason about high level architecture choices, how you decide upon solutions given various constraints, and overall problem solving skills.

Onsite: Join us in person to meet the rest of the team.

Reference checks: We will conduct a minimum of 2 reference checks. Be prepared to share references of prior employers, classmates, or anyone you might have worked with in the past.

We may use AI note takers during interviews to make sure we are fully focused on you. Please let us know if this is an issue for you.