Deep Learning Research Engineer Intern

DataRobot•Boston, MA

About The Position

DataRobot is developing a multi-target joint probabilistic foundation model intended to support temporal forecasting, unordered tabular regression and classification, missing-data completion, and mixed-modality inputs and outputs. The model aims to learn coherent joint structure across connected variables, rows, horizons, and scenarios, with a focus on practical value in real-world use cases as well as strong benchmark performance. This research-engineering internship is intended for someone operating at post-doctoral level or close to it who wants to shape the architecture of a probabilistic foundation model. The work centers on the model itself: encoders for temporal, tabular, and mixed-modality inputs; attention and sequence mechanisms; and distributional output heads and decoding schemes for coherent joint predictions. The intern will reason about architectural capabilities, implement and train PyTorch models, and evaluate them in a production-quality codebase. Stochastic modeling is a strong plus, particularly for contributing to synthetic data generation based on stochastic dynamics.

Requirements

Strong PyTorch skills and hands-on experience building and training deep models.
Solid understanding of Transformers, attention variants, and long-context or state-space sequence architectures, including tradeoffs among quality, memory, and latency.
Experience with probabilistic modeling in neural networks, such as distributional output heads, likelihood-based or proper-scoring-rule losses, mixture or flow models, or related uncertainty-aware learning setups.
Strong foundation in probability and statistics, including reasoning about densities and masses, dependence between variables, and calibration.
Ability to connect architectural ideas to working GPU-native implementations, controlled experiments, and diagnostics that explain why a change helped.
Strong engineering habits, including readable code, tests, reproducible experiments, and disciplined evaluation of model changes.
Ability to debug training instability, reason about changes in results, and iterate quickly from hypothesis to evidence.

Nice To Haves

Working knowledge of stochastic processes and stochastic differential equations, including drift and diffusion, jumps, regime switching, mean reversion, heavy tails, and numerical simulation of these dynamics.
Experience designing synthetic data generators or simulation-based training curricula, and understanding of how the training distribution shapes model learning.
Depth in a domain with rich stochastic structure, such as finance, energy, commodities, or a similarly quantitative field.
Familiarity with tabular or mixed-modality deep learning, including categorical, ordinal, count, and bounded targets alongside continuous targets.
Background in stochastic modeling.

Responsibilities

Design and improve core model architecture, including input encoders for temporal, unordered tabular, and mixed-modality data; attention factorizations across variables, rows, and horizons; and distributional output heads and decoding strategies.
Run controlled architecture studies, including ablations, scaling analyses, and memory and throughput profiling, and use results to inform design decisions.
Build scalable PyTorch implementations that support larger input and output spaces, improved throughput, and tighter memory budgets.
Study how architectural choices, data-generation choices, and inference constraints affect benchmark quality and real-world usefulness.
For candidates with a stochastic-modeling background, extend the synthetic-data engine with richer stochastic dynamics, constraints, dependence structures, heavy tails, and regime behavior.
Turn research ideas into robust implementations and credible empirical results.
Implement, train, and evaluate models, using controlled experiments and diagnostics to assess architectural changes.