Research Engineer - Optimization & Machine Learning

• About Smack
• Smack is building domain-specific models for the sole purpose of delivering AI-enabled decision-making for modern warfare. Our team has decades of military leadership experience at every level. We understand the problem better than anyone, and we know exactly what our end users need to solve it.

• About the Role
• Our scientists work on decision-making under uncertainty: large-scale optimization, multi-agent reinforcement learning, simulation, and more. We are hiring a Research Engineer to help investigate models written in pencil and convert them to code we can run on the stack.
• This is a research role, not a software engineering one, but an ability to orient within an evolving software stack and contribute high-quality code is a strict requirement.

• What You’ll Do
• Participate in scientific investigations end to end: form a hypothesis, run the investigation, return a report, and develop prototypes based on the investigation as applicable
• Translate math to code: a scientist drafts a model, you meet on it, and you produce a working implementation for evaluation, quickly and cleanly enough that a colleague could pick it up
• Assess methods from literature: inputs, outputs, compute requirements, and fit against our constraints
• Understand, maintain, and update installations of previously built models

• Required Skills
• There are two distinct backgrounds we’re hiring for. You only need one.

• For everyone
• 2+ years actively writing code in Python (research code counts), especially code that others have used, extended, or depended on
• Mathematical maturity: you can read a drafted model or a paper and understand it well enough to implement it faithfully

• If Optimization / Operations Research leaning
• Strong grasp of probabilistic models: comfort reasoning about random variables and uncertainty as they arise in modeling and analysis
• Working knowledge of mathematical programming and solvers (e.g., Gurobi, CPLEX)
• Familiarity with decision-making concepts such as Markov decision processes, partial observability, network design, and policy representation

• If Machine Learning leaning
• Working knowledge of ML, especially deep and reinforcement learning fundamentals
• Fluency with a modern framework (PyTorch or JAX) for building training and evaluation loops, reading and reproducing papers
• Current with the state of the art: you follow the literature and can speak to which recent methods matter and why

• Nice to Have
• MS or PhD in Mathematics, Computer Science, Operations Research, or a related quantitative field
• Research experience in mathematics or ML: designing, supporting, or reproducing technical work
• Exposure to generative methods like sequence, diffusion, and flow models

• Smack's Mission & Values
• Smack’s mission is to Deliver Decision Dominance to the Department of War, our allies, and partners.
• We’re here to serve: We work extremely hard every day in service of our mission and each other.
• We’re here to win: There are no participation trophies in war. We’re not satisfied with anything less than winning.
• We’re a culture of “And”: The Iron Triangle is for losers. Even the laws of physics can be bent with a little motivation.

Compensation:

Our full package includes base, equity, and performance incentives.