Applied Research Scientist III
Sphere Digital Recruitment Group
•
New York, NY
0
0
0
0
Mid-Senior level
Posted March 14, 2026
Job link
Thinking about this job
Responsibilities
Commitments
Responsibilities
- Develop and evaluate algorithms for real-time decisioning across bidding, pacing, dynamic allocation, traffic shaping, and marketplace optimization.
- Build statistics-driven, classical ML models grounded in probability, statistical learning, online learning, Bayesian methods, optimization, forecasting, and causal inference.
- Design interpretable, efficient algorithms optimized for large-scale, low-latency production environments.
- Rapidly prototype ideas and run experiments with fast feedback cycles, iterating based on real marketplace outcomes.
- Own scientific solutions end-to-end: problem formulation, mathematical modeling, algorithm design, experimentation, production deployment, and ongoing monitoring.
- Partner closely with engineering and product teams to deploy and refine algorithms in live systems where performance is measured directly by business impact.
- Contribute to internal scientific knowledge-sharing and, where appropriate, external research and publications.
- Identify new algorithmic opportunities across the exchange to improve efficiency, revenue, and overall system performance.
- Work with large-scale distributed systems and big-data platforms (e.g., Spark) to support model development and evaluation.
- Applied Data Scientist III - Requirements
Commitments
If you require any adjustments or additional support during the recruitment process for any reason whatsoever, please let us know.
Not Met Priorities
What still needs stronger evidence
Requirements
- Applied Data Scientist III - Requirements
- Open to fresh PhD graduates and candidates slightly below Senior level with strong fundamentals and research depth.
- Ideally ~5-7 years of experience, with flexibility for exceptional early-career PhDs; preference is not for overly senior or Staff-level profiles.
- Strong theoretical grounding in probability, statistics, algorithms, optimization, and statistical learning theory.
- Experience with classical machine learning models and simpler deep learning approaches commonly used in Ad Tech prior to the recent LLM wave.
- Proven ability to design original algorithms, not just apply existing libraries or pre-built models.
- Strong coding skills in Python, with experience in scientific computing frameworks (NumPy, SciPy, PyTorch, TensorFlow).
- Substantial experience working with large-scale data and distributed computing systems (e.g., Spark).
- Demonstrated ability to take research-grade models from prototype into production in high-scale, real-time systems.
- Deeply grounded in algorithmic research with a strong theoretical foundation.
- Experienced in reinforcement learning, marketplace modeling, and auction/optimization problems .
Preferred Skills
- PhD strongly preferred, particularly in Mathematics, Probability & Statistics, Machine Learning, Physics, or similarly rigorous quantitative fields.
- Open to fresh PhD graduates and candidates slightly below Senior level with strong fundamentals and research depth.
- Ideally ~5-7 years of experience, with flexibility for exceptional early-career PhDs; preference is not for overly senior or Staff-level profiles.
- Strong theoretical grounding in probability, statistics, algorithms, optimization, and statistical learning theory.
- Experience with classical machine learning models and simpler deep learning approaches commonly used in Ad Tech prior to the recent LLM wave.
- Proven ability to design original algorithms, not just apply existing libraries or pre-built models.
- Strong coding skills in Python, with experience in scientific computing frameworks (NumPy, SciPy, PyTorch, TensorFlow).
- Substantial experience working with large-scale data and distributed computing systems (e.g., Spark).
- Demonstrated ability to take research-grade models from prototype into production in high-scale, real-time systems.
- Publication record or meaningful research contributions (conference papers, workshops, or open-source work) is strongly preferred.
- Experience in marketplaces, auctions, or exchange systems is a plus, but auction theory and economics backgrounds are not required.
- Deeply grounded in algorithmic research with a strong theoretical foundation.
- Experienced in reinforcement learning, marketplace modeling, and auction/optimization problems .
- Hands-on scientists who enjoy bringing novel algorithms from research into high-impact production environments.
- Curious, rigorous thinkers who thrive in a fast-paced, research-driven culture.
- Motivated by solving some of the most complex, high-scale decisioning challenges in ad marketplaces.
- PhD in statistics or mathematics or computer science Mathematical foundations
Education
- (Not required) – PhD strongly preferred, particularly in Mathematics, Probability & Statistics, Machine Learning, Physics, or similarly rigorous quantitative fields.
- (Not required) – Open to fresh PhD graduates and candidates slightly below Senior level with strong fundamentals and research depth.
- (Not required) – Ideally ~5-7 years of experience, with flexibility for exceptional early-career PhDs; preference is not for overly senior or Staff-level profiles.
- (Not required) – PhD in statistics or mathematics or computer science Mathematical foundations
Title: Applied Scientist III Location: San Mateo, CA (Hybrid/Very Flexible) Industry: Advertising Services Salary: $160,000 - $220,000 base + RSUs + benefits Overview We have partnered with a global ad tech business to hire an Applied Data Scientist III for their algorithmic and research science team, which sits at the core of their Exchange business which is the real-time marketplace that powers ad auctions worldwide. This team is highly technical and deeply mathematical, with many scientists holding advanced quantitative PhDs. The group focuses on classical machine learning, probability, statistics, and algorithmic decision-making, building interpretable and efficient models that operate under tight latency constraints. This is a true algorithmic research + production science role, ideal for a Senior-level Data Scientist (or strong candidate one level below Senior, including fresh PhDs) who wants to grow within a technically rigorous environment. You will design and deploy algorithms directly into large-scale production systems that influence trillions of real-time decisions daily, with immediate and measurable business impact. Applied Data Scientist III - Responsibilities
Develop and evaluate algorithms for real-time decisioning across bidding, pacing, dynamic allocation, traffic shaping, and marketplace optimization.
Build statistics-driven, classical ML models grounded in probability, statistical learning, online learning, Bayesian methods, optimization, forecasting, and causal inference.
Design interpretable, efficient algorithms optimized for large-scale, low-latency production environments.
Rapidly prototype ideas and run experiments with fast feedback cycles, iterating based on real marketplace outcomes.
Own scientific solutions end-to-end: problem formulation, mathematical modeling, algorithm design, experimentation, production deployment, and ongoing monitoring.
Partner closely with engineering and product teams to deploy and refine algorithms in live systems where performance is measured directly by business impact.
Contribute to internal scientific knowledge-sharing and, where appropriate, external research and publications.
Identify new algorithmic opportunities across the exchange to improve efficiency, revenue, and overall system performance.
Work with large-scale distributed systems and big-data platforms (e.g., Spark) to support model development and evaluation. Applied Data Scientist III - Requirements
PhD strongly preferred, particularly in Mathematics, Probability & Statistics, Machine Learning, Physics, or similarly rigorous quantitative fields.
Open to fresh PhD graduates and candidates slightly below Senior level with strong fundamentals and research depth.
Ideally ~5-7 years of experience, with flexibility for exceptional early-career PhDs; preference is not for overly senior or Staff-level profiles.
Strong theoretical grounding in probability, statistics, algorithms, optimization, and statistical learning theory.
Experience with classical machine learning models and simpler deep learning approaches commonly used in Ad Tech prior to the recent LLM wave.
Proven ability to design original algorithms, not just apply existing libraries or pre-built models.
Strong coding skills in Python, with experience in scientific computing frameworks (NumPy, SciPy, PyTorch, TensorFlow).
Substantial experience working with large-scale data and distributed computing systems (e.g., Spark).
Demonstrated ability to take research-grade models from prototype into production in high-scale, real-time systems.
Publication record or meaningful research contributions (conference papers, workshops, or open-source work) is strongly preferred.
Experience in marketplaces, auctions, or exchange systems is a plus, but auction theory and economics backgrounds are not required. This role is ideal for candidates who are:
Deeply grounded in algorithmic research with a strong theoretical foundation.
Experienced in reinforcement learning, marketplace modeling, and auction/optimization problems .
Hands-on scientists who enjoy bringing novel algorithms from research into high-impact production environments.
Curious, rigorous thinkers who thrive in a fast-paced, research-driven culture.
Motivated by solving some of the most complex, high-scale decisioning challenges in ad marketplaces. Desired Skills and Experience
PhD in statistics or mathematics or computer science Mathematical foundations
Sphere Digital Recruitment currently have a variety of job opportunities across digital so feel free to get in touch with us to find out how we can help you. Please take a look at our website. Sphere is an equal opportunities employer. We encourage applications regardless of ethnic origin, race, religious beliefs, age, disability, gender or sexual orientation, and any other protected status as required by applicable law. If you require any adjustments or additional support during the recruitment process for any reason whatsoever, please let us know.
Develop and evaluate algorithms for real-time decisioning across bidding, pacing, dynamic allocation, traffic shaping, and marketplace optimization.
Build statistics-driven, classical ML models grounded in probability, statistical learning, online learning, Bayesian methods, optimization, forecasting, and causal inference.
Design interpretable, efficient algorithms optimized for large-scale, low-latency production environments.
Rapidly prototype ideas and run experiments with fast feedback cycles, iterating based on real marketplace outcomes.
Own scientific solutions end-to-end: problem formulation, mathematical modeling, algorithm design, experimentation, production deployment, and ongoing monitoring.
Partner closely with engineering and product teams to deploy and refine algorithms in live systems where performance is measured directly by business impact.
Contribute to internal scientific knowledge-sharing and, where appropriate, external research and publications.
Identify new algorithmic opportunities across the exchange to improve efficiency, revenue, and overall system performance.
Work with large-scale distributed systems and big-data platforms (e.g., Spark) to support model development and evaluation. Applied Data Scientist III - Requirements
PhD strongly preferred, particularly in Mathematics, Probability & Statistics, Machine Learning, Physics, or similarly rigorous quantitative fields.
Open to fresh PhD graduates and candidates slightly below Senior level with strong fundamentals and research depth.
Ideally ~5-7 years of experience, with flexibility for exceptional early-career PhDs; preference is not for overly senior or Staff-level profiles.
Strong theoretical grounding in probability, statistics, algorithms, optimization, and statistical learning theory.
Experience with classical machine learning models and simpler deep learning approaches commonly used in Ad Tech prior to the recent LLM wave.
Proven ability to design original algorithms, not just apply existing libraries or pre-built models.
Strong coding skills in Python, with experience in scientific computing frameworks (NumPy, SciPy, PyTorch, TensorFlow).
Substantial experience working with large-scale data and distributed computing systems (e.g., Spark).
Demonstrated ability to take research-grade models from prototype into production in high-scale, real-time systems.
Publication record or meaningful research contributions (conference papers, workshops, or open-source work) is strongly preferred.
Experience in marketplaces, auctions, or exchange systems is a plus, but auction theory and economics backgrounds are not required. This role is ideal for candidates who are:
Deeply grounded in algorithmic research with a strong theoretical foundation.
Experienced in reinforcement learning, marketplace modeling, and auction/optimization problems .
Hands-on scientists who enjoy bringing novel algorithms from research into high-impact production environments.
Curious, rigorous thinkers who thrive in a fast-paced, research-driven culture.
Motivated by solving some of the most complex, high-scale decisioning challenges in ad marketplaces. Desired Skills and Experience
PhD in statistics or mathematics or computer science Mathematical foundations
Sphere Digital Recruitment currently have a variety of job opportunities across digital so feel free to get in touch with us to find out how we can help you. Please take a look at our website. Sphere is an equal opportunities employer. We encourage applications regardless of ethnic origin, race, religious beliefs, age, disability, gender or sexual orientation, and any other protected status as required by applicable law. If you require any adjustments or additional support during the recruitment process for any reason whatsoever, please let us know.