Build mathematical and algorithmic models that represent real-world systems, and use these models to generate realistic synthetic data and decision-support algorithms for Eskwelabs. Fellows in this track focus on abstracting messy real-world processes into structured assumptions, equations, and simulations—then translating those models into working Python code. AI tools are used to accelerate domain research, model planning, and coding, enabling fellows to produce outputs that match professional data science standards.

Your work will contribute to the following outcomes:
‍1. High-quality synthetic datasets that realistically mirror real-world systems (e.g. recruitment funnels, learning journeys, organizational processes), enabling experimentation, demos, testing, and training without relying on sensitive or unavailable real data.
‍
2. Actionable internal algorithms and models (e.g. allocation, clustering, recommendation) that support better operational and strategic decisions at Eskwelabs.

3. Reusable modeling frameworks and notebooks that document assumptions, formulas, and logic—allowing future analysts, AI agents, or fellows to extend or adapt the models.

4. Stronger modeling culture at Eskwelabs, where decisions and simulations are grounded in explicit assumptions, testable logic, and transparent trade-offs

A. Synthetic Data Generation
This task centers on modeling a real-world system and using that model to generate realistic synthetic data. Fellows must rapidly acquire domain understanding, identify key variables and relationships, and encode these relationships mathematically before generating data. Importantly, some variables may be modeled internally but omitted from the final dataset to preserve realism and complexity.

‍Example: Given the topic “recruitment funnel,” build a generative model that simulates applicants, screening stages, interviewer bias, role constraints, and acceptance decisions—then output a synthetic dataset that preserves realistic correlations, drop-off rates, and edge cases.
‍
‍B. Algorithms & Decision Models
This task involves designing and implementing algorithms that support internal decisions. Fellows translate business or operational problems into formal problem statements, define objective functions and constraints, and implement models using Python.

‍Example: Build a resource allocation algorithm that assigns mentors to cohorts based on availability, specialization, and load-balancing constraints, or a recommendation model that suggests learning pathways based on learner behavior.

This is a task-based, production-oriented modeling role focused on abstraction, simulation, and algorithm design. Work arrives as problem statements with a real-world context and desired outcome. Fellows are expected to reason from first principles, formalize assumptions, and implement models in Python notebooks. AI tools are used as productivity multipliers—for rapid domain research, model sketching, code generation, and validation—but final outputs require human judgment, mathematical rigor, and careful testing. Success is measured by realism, correctness, clarity of assumptions, and usefulness of the resulting models.

This track is ideal for individuals who naturally think about the world in terms of systems, variables, and equations—and enjoy turning ambiguity into formal structure.

You might come from backgrounds such as:
- Data Science & Quantitative Fields: statistics, data science, applied math, operations research.
- Engineering & Computing: computer science, engineering, physics, or related technical fields.
- Economics & Social Science (Quantitative): economics, quantitative psychology, or policy fields with modeling exposure.

What matters most is not mastery, but mindset. You should enjoy asking questions like “What is the underlying process here?”, “What variables matter?”, and “How would this behave if conditions changed?” This is a specialist-leaning role for fellows who want to deepen technical and modeling skills.

- Translating real-world systems into formal models and simulations
- Probabilistic thinking and synthetic data generation
- Algorithm design (optimization, clustering, recommendation)
- Python-based modeling in Jupyter notebooks
- Assumption-driven reasoning and model validation
- Using AI for rapid domain learning, code scaffolding, and iteration

Traits
- Strong abstraction ability; naturally models real-world phenomena mathematically
- Comfort with ambiguity and incomplete information
- Curiosity and speed in acquiring new domain expertise
- High bar for logical consistency and internal validity
- Willingness to document assumptions and limitations clearly

‍Skills
- Python for data science (NumPy, pandas, basic modeling libraries)
- Mathematical reasoning (probability, basic optimization, linear algebra concepts)
- Algorithmic thinking and decomposition
- AI usage for research, model planning, and coding acceleration
- Notebook-based experimentation and iteration

- Data Scientist / Applied Scientist: modeling, simulation, and decision support

- Machine Learning Engineer (Model-Focused): feature logic, training data design

- Operations Research / Optimization Specialist: allocation and constraint-based systems

- Analytics Engineer / Modeling Lead: building reusable analytical frameworks

- AI Product or Systems Designer: defining how models power real-world tools

All modeling tasks are listed in a shared task tracker with a problem statement, expected output, deadline, and submission format. Fellows claim tasks, work independently in Python notebooks, and submit code, outputs, and documentation by the stated deadline.

Fellows are expected to communicate proactively if assumptions, constraints, or data requirements are unclear. Submissions typically include a notebook, generated datasets or model outputs, and a short written explanation of assumptions and results. Tasks are reviewed and marked as Approved or Revisions Needed.

You will report to the Data and Decision Systems lead of Eskwelabs. They meet with fellows twice a month to review models, challenge assumptions, and demonstrate better modeling patterns, especially around synthetic data realism, algorithm design, and AI-assisted workflows.

You will be judged on the quality of your models, the realism of synthetic data, and the clarity of your assumptions and documentation. Strong performance includes clean, reproducible notebooks; thoughtful modeling choices; and outputs that meaningfully represent real-world systems rather than simplistic toy examples.