Staff Machine Learning Engineer, Pinterest
Content Recommendations
More information about location
Visual bookmarking tool for saving & discovering creative ideas
Be an early applicant
Visual bookmarking tool for saving & discovering creative ideas
1001+ employees
Be an early applicant
More information about location
1001+ employees
Company mission
To bring everyone the inspiration to create a life they love.
Role
Who you are
- 5+ years building ML products end‑to‑end, including 2+ years as a tech lead driving multi‑quarter roadmaps and cross‑functional execution
- Deep hands‑on experience with NLP/LLM training and inference (PyTorch, Python); strong grounding in evaluation, prompt/data design, and fine‑tuning
- Proven track record shipping models at scale: feature/data pipelines, online serving, monitoring/observability, and cost/perf trade‑offs
- Strong software engineering in Python with an eye for software engineering best practices
- Experience mentoring senior engineers and influencing partner teams
- Masters or PhD in ML related studies
- LLM efficiency techniques (LoRA/adapters, distillation, quantization, prompt caching) and cost control strategies
- MLOps at scale with tools like Airflow, Spark/Presto, Triton, vLLM
What the job involves
- Content Mining identifies the best sources to acquire content for Pinterest (websites, merchants, social accounts), optimizes how we acquire it, and extracts structured attributes from that content at high scale. Our work powers inspiring, accurate, and engaging Pins
- This is a high performing end to end ML team, with a recent paper in KDD: “Cross-Domain Web Information Extraction at Pinterest”
- We’re hiring a Staff ML Engineer to serve as the technical lead for a 5‑engineer team (4 MLEs + you)
- As a tech lead, you will define the multi‑quarter technical vision and roadmap, lead execution and mentor engineers
- The majority of your time will be spent with hands‑on designing, training, and shipping ML systems—especially LLM/NLP models for extraction and
- Own the long‑term architecture, roadmap, and execution for source discovery, acquisition optimization, and content understanding
- Lead design reviews, set engineering standards, and drive cross‑team alignment with Product, Data, and Infra
- Mentor and uplevel MLEs through technical direction, pairing, and reviews
- Train/fine‑tune LLMs and NLP models for classification, extraction, and instruction‑following; design eval loops and guardrails
- Design features and frameworks for sharing features across models
- Productionize models for large-scale inference; drive latency, reliability, and cost efficiency (quantization, distillation, caching)
- Establish offline/online evaluation, gold sets, and automated regressions; run A/B and canary/shadow launches
- Work with human and automated labeling sources to define data labeling standards
- Partner on data strategy, labeling/weak supervision, and feedback loops to expand coverage and improve precision/recall
- Define and meet SLOs for data quality, model performance, and serving reliability; lead incident playbooks and postmortems
- Measure and drive downstream impact on revenue and engagement
Our take
Inspiration and ideas can pop into your head at any time, and with the growth of social media and networking, sharing these interests and gaining exposure has never been easier. Pinterest exists to display these moments visually, as a kind of digital pinboard allowing users to save and share digital images which have inspired them.
Pinterest's emphasis on visual discovery has differentiated it from competitors and created a more positive and fulfilling experience for its users. Pinterest in many ways provides a respite from the ceaseless negativity that plagues social media such as Facebook and Twitter, and this is perhaps responsible for its userbase expanding to 450 million users in 2022, driven by uptake mostly outside of North America.
Pinterest has not been immune to woes affecting the global tech sector and made about 5% of its workforce redundant at the end of 2022 and the start of 2023. Having restructured to face a challenging climate, it is well-positioned to resume its growth, which has increased consistently year on year. Its immediate goals are to continue to grow its user base and expand the uptake of its short video product, which it aims to monetize.
Kirsty
Company Specialist at Welcome to the Jungle
Insights
9% employee growth in 12 months
Company
Company HQ
SoMa, San Francisco, CA
Leadership
Ben Silbermann
(CEO)Worked as a Product Specialist at Google for 2 years before leaving to work on his own apps, which led to founding Pinterest
Salary benchmarks
We don't have enough data yet to provide salary benchmarks for this role.
Submit your salary to help other candidates with crowdsourced salary estimates.
Share this job
View 74 more jobs at Pinterest