Senior Data Scientist: AI Training Data (2-4 Months Contract)

Company: BespokeLabs (VC-backed; founded by IIT & Ivy League alumni)

Location: Remote

Role Type: Contract (2-4 Months)

Time Commitment: 40 hrs/week (Full-time availability required)

Compensation: Hyper-competitive hourly rate (matching top-tier Senior Data Scientist bands) Experience: 6+ years

About BespokeLabs

BespokeLabs is a premier, VC-backed AI Research lab with an exceptionally talent-dense team of IIT and Ivy League alumni. We don’t just build tooling around AI—we build the massive-scale data systems and reasoning architectures that directly power next-generation models. Our research shapes the frontier of AI: we’ve published breakthroughs like GEPA, driven foundational datasets like OpenThoughts, and shipped state-of-the-art models including Bespoke-MiniCheck and Bespoke-MiniChart. More on our website bespokelabs.ai :)

Role Overview

We are looking for a high-impact Senior Data Scientist for an intensive, 2-month sprint. You will leverage your deep expertise in production-grade machine learning and applied statistics to develop the algorithms and logic that curate and evaluate datasets for advanced AI model training.

This is not a traditional model-building or research role. We need a seasoned practitioner who has already owned the end-to-end DS lifecycle at scale. You will use your intuition for feature engineering, statistical validity, and large-scale data processing to programmatically generate, shape, and validate AI training data.

What You Will Do (The Contract)

* Algorithm Design: Design and implement custom statistical models and programmatic logic (e.g., anomaly detection, active learning, similarity scoring) to evaluate data quality, complexity, and redundancy at scale.

* Hands-on At-Scale Coding: Write scalable PySpark and Python (NumPy/Pandas) code to apply these algorithms across massive datasets, translating experimental logic into reliable, large-scale workflows.

* Metric Formulation: Develop custom quantitative metrics and heuristic benchmarks to rigorously assess the fidelity and suitability of data subsets for specific AI training objectives.

* Validation & Iteration: Run high-speed validation cycles, analyzing the output of data-curation algorithms to diagnose skew, bias, or noise, and iteratively refining the logic.

* High-Level Curation: Apply Senior-level domain expertise in predictive modeling and feature engineering to ensure the final training inputs meet the strict standards required for state-of-the-art ML systems.

What You Bring to the Table (Your Past Experience)

To be successful in this contract, you must have a track record of:

* The End-to-End DS Lifecycle: Framing problems, modeling, validation, production, and iteration.

* Production Ownership: Building and deploying ML and statistical models on large-scale datasets.

* Large-Scale Data Processing: Working with Apache Spark to develop scalable feature pipelines and offline training workflows.

* Experimentation: Designing and analyzing rigorous experiments (A/B tests, causal inference).

* Impact: Translating complex model outputs into clear product and business decisions.

Required Qualifications (Non-Negotiable)

* Experience: 6+ years as a Data Scientist or Applied Scientist.

* Production Background: Proven ownership of models running in production environments.

* Applied Statistics: Strong background in applied statistics and experimentation frameworks.

Core Technical Skills

* Languages: Python (NumPy, Pandas, Scikit-learn, PyTorch / TensorFlow) and Strong SQL.

* Big Data: Apache Spark (PySpark or Spark SQL) for large-scale data processing.

* Methodologies: Feature engineering, model evaluation, statistical modeling, and hypothesis testing.

Strong Signals (Highly Valued)

* Scale: Models trained on TB-scale datasets.

* Domain Specificity: Experience in high-complexity domains such as: Recommendations, Pricing, Fraud / risk, Search / ranking, or Growth & experimentation.

* Collaboration: Experience deploying models alongside data engineering pipelines.

Out of Scope (Who Should Not Apply)

* BI / reporting-only roles

* SQL-only analysts

* Research-only ML roles with no production ownership

* Early-career profiles