AI​/ML Engineer, RL Environments - Asset Hub

Company Overview

Most companies don’t end in acquisition — they end in closure. Yet shutdown has never been given the same rigor, clarity, or professionalism as the moments that came before it. Nearly 4 million companies shut down every year, creating a massive, overlooked market. Simple Closure is building the infrastructure to change that.

Having raised over $20M from leading VCs and trusted by more than 7,000 companies to date, we streamline the entire business dissolution process — from legal and tax to operational and administrative — as one coordinated effort instead of a patchwork of disconnected steps. We go a step further, helping companies sell their remaining assets and return the most value to their stakeholders, through our Asset Hub.

As the leading platform for company shutdowns, we have something no one else does: a continuous, proprietary flow of companies winding down, and the assets they built along the way. Our buyers span the top AI labs, RL environment providers, vertical and enterprise agent builders, VCs and deal‑flow partners, domain marketplaces, and much more. Dissolution is our wedge; the marketplace is a big part of where we’re headed.

Joining our Asset Hub team means helping to define a new category from the ground up, working on complex, high‑stakes problems, and building the products and processes that guide founders through one of the most consequential moments in a company’s life.

Simple Closure is seeking an AI/ML Engineer to turn Asset Hub’s one‑of‑a‑kind inventory, and existing AI buyer relationships, into high‑value AI‑training products. When companies shut down, we acquire the real assets they built — production codebases, work spaces, and databases. Your job is to find the opportunities hidden in that inventory and build them: transforming real‑world assets into derivative works — reinforcement‑learning environments, agentic task suites, evaluations and verifiers, and training datasets — that are far more valuable to the AI labs, RL‑environment providers, and agent builders who already buy from us.

This is a hands‑on building role for someone who comes from the RL‑environments and AI‑training world and has actually created environments and products used to train or evaluate models — that background is essential. You’ll prototype fast, then harden what works into repeatable pipelines, working in a small, dedicated Asset Hub pod alongside product, engineering, and the GM of Asset Hub.

Candidates MUST be located in the New York City Metro area.

• Take Asset Hub’s unique real‑world assets — production codebases, work spaces, and databases — and identify how each can become a high‑value AI‑training product: RL environments, agentic task suites, evals and verifiers, benchmarks, and fine‑tuning or trajectory datasets.

• Design and build the pipeline that turns a raw asset into a derivative work: repository ingestion, test harnessing, commit‑mining for task extraction, Docker/sandbox reproducibility, verifier and reward scripts, and QA tooling.

• Wrap real data in interactive environments — sandboxed application state, MCP servers, and browser/Playwright layers — that buyers can train and evaluate agents against.

• Spot the commercial opportunity in the inventory: which assets map to current lab and RLE demand, and what derivative product maximizes their value.

• Prototype quickly, then harden the best ideas into repeatable, scalable pipelines so derivative‑work creation isn’t one‑off.

• Partner with the Asset Hub buyer/BD side and directly with technical stakeholders at labs and RLE buyers to shape what we build to their training needs.

• Work with sensitive material — codebases, workspace exports, and proprietary datasets — with strong attention to security, privacy, licensing, and PII handling.

• Write clean, well‑tested code and use AI tooling to move faster; collaborate closely with product, engineering, and the GM of Asset Hub.

• RL‑environments / AI‑training background (critical): you’ve built RL environments and/or products used to train or evaluate models — environments, agentic task suites, evals, benchmarks, or…