SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Salary: USD 140,000 - 200,000 / annual
SimpleClosure is building infrastructure for company shutdowns and dissolution, serving a market of nearly 4 million companies annually. The company has raised over $20M and is trusted by 7,000+ companies. Beyond streamlining the dissolution process itself, SimpleClosure's Asset Hub acquires real assets from shuttering companies—production codebases, workspaces, databases—and transforms them into high-value AI-training products.
The AI/ML Engineer role is a hands-on position focused on turning Asset Hub's unique inventory into derivative works for the AI training ecosystem: reinforcement-learning environments, agentic task suites, evaluations and verifiers, training datasets, and benchmarks. You'll work in a small, dedicated pod alongside product, engineering, and the Asset Hub GM.
Key responsibilities include identifying how raw assets can become high-value AI-training products; designing and building pipelines that ingest repositories, harness tests, extract tasks from commit history, ensure Docker/sandbox reproducibility, and create verifier and reward scripts; wrapping real data in interactive sandboxed environments (application state, MCP servers, browser/Playwright layers) for agent training and evaluation; spotting commercial opportunities in the inventory by mapping assets to current lab and RLE demand; prototyping quickly and hardening successful ideas into repeatable, scalable pipelines; partnering with buyers and technical stakeholders at labs to shape products to their training needs; and handling sensitive material (codebases, workspace exports, proprietary datasets) with strong attention to security, privacy, licensing, and PII.
The ideal candidate has 4–8 years of engineering experience with meaningful time in RL environments, AI-training data, or model evaluation (at a lab, RLE/eval company, or team that shipped training environments). Critical skills include strong Python, Docker, and CI/test infrastructure; ability to build reproducible sandboxes from messy real-world code; familiarity with LLM evaluation and agent harnesses (SWE-bench-style setups, Verifiers, HUD); verifier/reward design including resistance to reward hacking; and a builder's temperament that moves projects from concept to production. Clear communication with lab and RLE researchers is essential. Nice-to-haves include contributions to public benchmarks or eval frameworks, post-training/fine-tuning data experience, and simulation or frontend skills (MCP, Playwright). A Bachelor's or Master's in Computer Science, Machine Learning, or related field (or equivalent practical experience) is expected. Team management experience is a plus.
