Senior Machine Learning Engineer

beehiiv is the all-in-one newsletter and creator platform powering the next generation of media companies and independent voices. We help creators and businesses launch, grow, and monetize newsletters—now serving tens of thousands of publishers and reaching hundreds of millions of readers monthly. With $20M+ ARR, 100% YoY growth, and a newly expanded product suite (ad network, website builder, paid tiers), we’re entering our next phase of hypergrowth.

We are seeking a motivated Senior Machine Learning Engineer to design, build, deploy, and own production ML systems that power core parts of beehiiv’s ad network, content intelligence, and growth ecosystem.

You’ll work across the entire ML lifecycle—problem framing, feature engineering, large-scale data processing, model training, evaluation, deployment, monitoring—and partner closely with Data Engineering, Product, and Platform teams.

This role is ideal for someone who wants to work on real-time optimization, high-dimensional modeling, vector embeddings, and ML-driven automation, all while shipping to production continuously.

The successful candidate will likely not have expertise across all of these areas, but rather deeper mastery in one of them (Data Science, Research & Development, MLOps, ML Engineering, etc.) with passable knowledge of the others. We value your unique perspective, and the day-to-day is highly dependent on the skills you bring.

Ideally, you have:

• 4–7+ years building ML systems end-to-end in production
• Deep experience with Python and common ML Libraries (pandas, NumPy, PyTorch, XGBoost, scikit-learn)
• Strong understanding of supervised learning, ranking models, anomaly detection, embeddings, optimization, and feature engineering
• Strong SQL and performance instincts (ClickHouse experience a huge plus)
• Ability to analyze extremely large datasets (potentially billions of events) efficiently
• Practical knowledge of statistics, experimental design, and investigative analytics
• Knowledge of the NLP ecosystem (Huggingface, SentenceTransformers, etc) and underlining Transformers architecture
• Familiarity with Airflow or similar orchestration tools
  

Nice-to-Have:

• Experience building models around digital ads, marketplaces, recommender systems
• Experience with entity resolution, especially in graph databases (Neo4j)
• Experience with vector databases (Chroma, Pinecone, Qdrant, etc.)
• Experience shipping ML components to production on Cloud through Docker, ECS, EKS
• Expertise in MLOps (MLFlow, Kubeflow, Feature Stores, etc), double points for experience in setup and maintenance on AWS
• Experience with single GPU inference systems (CUDA, MPS)
• Experience with multi-objective optimization or constrained solvers
• Prior startup experience or ownership-heavy engineering roles

What you'll be responsible for: 

Your core responsibilities will center on our ad network and high-volume, real-time systems. 

Ad Network Optimization:

• Build and improve models for CTR prediction, publisher ad acceptance probability, publisher–advertiser matching, content quality classifiers, ad creative optimization, etc
• Develop multi-stage optimization systems that assign ad opportunities using xarray tensors, constrained solvers, and multi-objective reward functions
• Improve inventory estimators, cold-start models, and dynamic feature pipelines
• Maintain and improve sentence-transformer embedding pipelines for publications, tags, campaigns, advertisers, and user behavior
• Own vector similarity search, clustering, topic modeling, and publication-level semantic analytics
• Personalization algorithms to optimize publisher experience within the ad network interface
• Creating lightweight microservices on AWS for model inference
• Developing and hosting APIs using Python FastAPI
• Research and implementation of new methods, models, and frameworks as the complexity of the ad network increases

Depending on your expertise and company priorities, you will have opportunities to contribute to the following initiatives.

ML Platform Engineering

• Design reusable training/serving pipelines using PyTorch, XGBoost, Airflow, Kubernetes/EKS, Docker, and vector databases
• Design MLOps architecture for artifact versioning and storage, back testing, ML centric CICD pipelines, etc
• Improve offline evaluation: log-loss, WMAPE, ROC AUC, calibration curves, multi-class softmax metrics
• Instrument production models with monitoring, drift detection, and alerting
  

Subscriber Profiles

• Entity resolution of subscriber-level data based on advertiser pixel tracking.
• Creating a dense representation of subscribers and publications behind basic content categorization that can be used in downstream models
• Advanced segment cohorts for better ad targeting and publisher reporting
• Hydrate profiles with external data sources to create stronger representations
  
  

Why this role might be for you:

• Join a fast-growing new team that prioritizes velocity 
• High impact and high visibility 
• You can take loosely defined objectives and develop appropriate proposals and solutions
• You have or are excited to work in a fast-paced startup environment.
• You’re excited about conducting research and experimenting with new technologies
• You can disagree and commit
  

Why this role might not be for you:

• Not a traditional 9-to-5 position.
• Need to manage multiple priorities and stakeholders.
• Must be comfortable with rapid iteration and frequent context switching.
• High expectations for both speed and quality.
• Must navigate complex organizational dynamics.

Why beehiiv?

• Bias towards action: Our first impulse is to act. We don’t get bogged down in unnecessary processes or bullsh**t. Perfection is the enemy of progress.
• Ownership mentality: This company is ours. We go the extra mile because that’s what owners do. Every day, people step up to take on tasks outside of their responsibilities and do whatever it takes for us to succeed.
• Building is in our DNA: We are obsessed with improving every aspect of our platform (and ourselves). Whether it’s our product, support, or partnerships, we never stop working to improve it.
• We answer to our users: Nothing matters more than serving our users. If our users fail, we fail.
• Ego comes second, but winning comes first: We put our egos aside to work collaboratively and build something special. It doesn't matter who's idea or who's vision, we're here to create the best outcome. We're here to win.
  

We'll take care of you:

• Competitive salary
• Stock Options
• Health, Dental, and Vision Insurance
• 401(k) employer match
• Unlimited PTO (mandatory 10 days per year minimum)
• Annual in-person team retreat
• Unlimited book budget
• Monthly Wellness Days (every third Friday of the month)