Algorithms and Theory Researcher (Research Coding Tasks)

What You'll Do8

• Turn a published paper that has released code into a research coding task an AI agent can attempt.
• Write the task statement that spells out the problem the agent must solve.
• Cut the code implementing the paper's core contribution and wrap the remaining pieces as scaffolding.
• Define held-out evaluation workloads and set a pass threshold for each metric.
• Build a hidden reference solution from the authors' original results.
• Add a second solution that clears the threshold to show the bar is reachable.
• Review agent attempts on your task and judge whether each one succeeds.
• Keep the benchmark setup reproducible so another person can rerun the evaluation.

Requirements6

• Published research at SODA, STOC, FOCS, ESA, ICALP or a venue of similar standing.
• A public code release for at least one paper, shared through GitHub or an artifact link.
• PhD in computer science or a related field, completed or in progress.
• First-author papers that report measured gains over a baseline.
• Comfort with Docker and benchmark setups others can reproduce.
• Depth in graph, approximation, streaming or online algorithms, backed by reference implementations.

Who Should Apply

A PhD-level algorithms researcher who publishes at theory venues and also ships code fits this work well. Expect to lean on a first-author paper with a measured gain over a baseline, plus Docker and benchmark setups another person can rerun.

Salary Insight