Simulation Engineer

Aaru•New York, NY

17h•Onsite

About The Position

Simulation Engineering owns the path from research idea to production system. Researchers prove out how to synthesize an audience, model a world, predict responses, score accuracy, or ship a result; you turn those ideas into robust, reusable, fast code — the objects, contracts, evaluations, and workflows that produce the product.

Requirements

Deep expertise in ML and AI, with meaningful prior work using Large Language Models in production systems
Designed and run rigorous ML experiments, moving from hypothesis to evaluation to deployment
3+ years of hands-on experience building ML/AI systems (research or applied)
Comfortable working across the full model lifecycle: data collection, training, evaluation, and production integration
Experience working cross-functionally with engineering and product teams to ship ML-powered features
Deeply curious and motivated to push the boundaries of what simulation technology can do

Nice To Haves

Publications at top ML venues (NeurIPS, ICML, ICLR, ACL)
Experience building or fine-tuning multi-agent systems
Worked on probabilistic modeling, Bayesian inference, or causal reasoning
Familiarity with our stack or demonstrate the ability to learn unfamiliar technologies quickly
Direct experience with simulation, agent-based modeling, Monte Carlo methods, or other modeling-and-prediction systems.
Experience with hybrid architectures that combine LLMs with classical statistical or ML methods.
Background in a quantitative or data-heavy domains — quant finance, computational social science, ad measurement, forecasting, statistics, or psychometrics.
A research sensibility: comfort reading papers, prototyping alongside researchers, and contributing to or co-authoring published work.

Responsibilities

Productionize the core simulation loop — audience generation, world modeling, response prediction.
Design the reusable abstractions that the product is made of: agent and population objects, simulation contracts and interfaces, evaluation harnesses, etc.
Build and own evaluation and accuracy infrastructure.
Make simulations fast and cheap enough to run at scale.
Partner closely with Simulation Research to harden hybrid LLM + classical architectures.
Build the workflows and tooling that let deployment and forward-deployed teams stand up new client simulations.
Create the calculation, analysis, and publication layers that turn raw agent output into the decision-ready artifacts customers see.
Hold the line on engineering quality across a fast-moving codebase.