Senior Data Scientist

Zefr

New York, US

On-site

Who this role is best for

Aimed at senior data scientists with hands-on LLM fine-tuning experience who thrive in multimodal AI research and production environments.

Best fit for

Specialists in optimizing LLM inference for real-world deployment at scale.
— “Experience with LLM inference optimization — KV cache management, quantization, multi-node GPU serving”
Researchers who bridge the gap between prototyping and production-grade systems.
— “Track record of taking ML models from research to production at scale”
Technologists comfortable with rapidly evolving foundation model ecosystems.
— “keep up with the rapidly evolving landscape of foundation models”

Things to consider

Must handle multi-terabyte datasets from multiple social platforms.
— “multi-terabytes of social media platform data from TikTok, YouTube, Facebook, Instagram, and Snap”
Role involves balancing quality, latency and cost tradeoffs.
— “balancing quality, latency and cost”

How to stand out

Document specific instances where you optimized inference costs versus accuracy.
— “Familiarity with cost/quality tradeoff analysis across LLM providers”
Highlight experience with compound AI systems combining multiple modalities.
— “build sophisticated compound AI systems that combine multiple models and modalities”
Showcase contributions to open-source LLM frameworks or optimizations.
— “Hands-on experience fine-tuning large language models (open-source and/or closed-source)”

Pace · Fast PacedCollaboration · MediumAutonomy · HighDecision Impact · TeamLevel · Senior

Derived from job-description analysis by Serendipath's career intelligence engine.

What success looks like

fine-tuning large language models
deploying llms at scale
balancing quality, latency and cost in llm inference

Typical background

bachelor's or master's degree in computer science or related field4+ years of professional experience in machine learning or data science

Skills & requirements

Required

Machine-learningPythonSQLLlm-inference-optimizationMultimodal-modelsDistributed-systemsGpu-accelerated-workloads

Preferred

Autonomous-coding-agentsInvestran

Stack & domain

PythonSQLSnowflakeQdrantRayPandasdbtFastapiAirflowAstronomerDbosGithub ActionsDockerTerraformKubernetesArgocdAWSGCPDatadogPyTorchTriton Inference ServerVllmNvidia DynamoNvidia Tensorrt LlmWeights And BiasesTransformersOnnxNvidia TensorrtDvcHuggingfaceCudaBaseten InferenceVoxel51Claude CodeCursorLeadershipCommunicationAIMachine LearningData Science

About the role

As a Senior Data Scientist at Zefr, you'll be at the forefront of developing and optimizing large language models for social media content analysis, requiring a blend of technical expertise and a passion for pushing the boundaries of AI.

Original posting from Zefr via Ashby

WHAT WE DO:
Zefr is the global leader in brand suitability targeting and measurement across the world's largest platforms. Zefr’s technology is helping to power the age of responsible marketing by putting advertisers in control of their content adjacencies based on their own unique brand safety and suitability preferences, mapped to the Global Alliance of Responsible Media's (GARM) industry standards. As an official YouTube Measurement Program Partner, Meta for Business Partner, and TikTok for Business Partner, the company leverages patented machine learning and AI technology (Cognition AI) to offer brands and agencies more precise and transparent brand safety and suitability activation and measurement solutions on scaled platforms. The company is headquartered in Los Angeles, California, with additional locations across the globe.
WHAT YOU’LL DO:
We are hiring a Senior Data Scientist focused on the research and productionalization of Large Language Models for multimodal social media content understanding. We work with multi-terabytes of social media platform data from TikTok, YouTube, Facebook, Instagram, and Snap. In this role you will fine-tune, optimize, and deploy LLMs that understand what hundreds of millions of videos, images, and text posts are about.
Your work will span the full lifecycle — from research and prototyping to production-grade serving at scale. You will fine-tune LLMs and perform inference optimizations, balancing quality, latency and cost. You will build sophisticated compound AI systems that combine multiple models and modalities into reliable, scalable pipelines.
We are excited to welcome someone who is passionate about pushing the boundaries of what LLMs can do and who can keep up with the rapidly evolving landscape of foundation models and inference infrastructure. This is a role where we both expect to learn from you and have you learn from us.
Tech stack:
Languages: Python, SQL
Data Stores: Snowflake, Qdrant
Data Processing: Ray, Pandas, DBT, FastAPI, Airflow, Astronomer, DBOS
DevOps: Github Actions, Docker, Terraform, Kubernetes, ArgoCD, AWS, GCP, Datadog
MLOps & Inference: PyTorch, Triton Inference Server, vLLM, Nvidia Dynamo, Nvidia TensorRT LLM, Weights and Biases, Transformers, ONNX, Nvidia TensorRT, DVC, HuggingFace, CUDA, Baseten Inference
Tools: Voxel51, Claude Code, Cursor
WHAT WE’RE LOOKING FOR:
Bachelor's or Master's degree in Computer Science or related field with 4+ years of professional experience in machine learning or data science
Hands-on experience fine-tuning large language models (open-source and/or closed-source)
Experience with LLM inference optimization — KV cache management, quantization, multi-node GPU serving, batching strategies, and serving frameworks (vLLM, TensorRT LLM, SGLang)
Track record of taking ML models from research to production at scale
Fluency with Python and SQL (specifically Snowflake)
Experience working with multimodal models and features
Familiarity with cost/quality tradeoff analysis across LLM providers and self-hosted models
Experience with distributed systems and GPU-accelerated workloads
Strong foundation in data structures, algorithms, and software design
Thorough testing and code review standards/practices
Strong verbal and written communication skills
Experience working with Autonomous Coding Agents
Openness to new technologies and creative solutions
BENEFITS (FOR US BASED EMPLOYEES):
Flexible PTO
Medical, dental, and vision insurance with FSA options
Company-paid life insurance
Paid parental leave
401(k) with company match
Professional development opportunities
13 paid holidays off
In-office, hybrid, and fully-remote work options available
“Summer Fridays” (shorter work days on select Fridays during the summertime)
In-office lunches and lots of free food
Optional in-person and virtual events (we like to celebrate!)
COMPENSATION (FOR US BASED EMPLOYEES):
The anticipated base salary for this position is between $200,000 and $225,000. Within the range, individual pay is determined by factors such as job-related skills, experience, and relevant education or training. If your compensation expectations fall outside of this range, it may still be worth having a conversation.LGBTQIA+ individuals, persons with disabilities, members of ethnic minorities, foreign-born residents, and veterans to apply even if you do not meet 100% of the qualifications.
Zefr is an equal opportunity employer that embraces diversity and inclusion in the workplace. We are committed to building a team that represents a variety of backgrounds, skills, and perspectives because we know this only makes us better. We strongly encourage women, persons of color, LGBTQIA+ individuals, persons with disabilities, members of ethnic minorities, foreign-born residents, and veterans to apply even if you do not meet 100% of the qualifications.

Source: Zefr careers (Ashby)