Research Engineer

Percepta•Boston, MA

22h•$180,000 - $400,000•Onsite

About The Position

As a Research Engineer at Percepta, you will work closely with the core research and product teams to build and scale modeling capabilities. The role focuses on the shared foundations supporting Percepta’s research tracks—world modeling, forecasting, user modeling, and optimization—including large-scale training and inference infrastructure and reliable distributed systems. You will build and own this foundation, work directly within customer environments to understand how it needs to adapt, and deliver production systems that turn research ideas into deployed solutions. The role combines infrastructure and systems engineering with rapid prototyping and customer-embedded applied AI work. You will contribute reusable capabilities to Mosaic, Percepta’s in-house toolkit, and help translate customer learnings into playbooks for future engagements. Percepta describes itself as an office-first team that values in-person collaboration and autonomy, with a focus on outcomes and customer success.

Requirements

Experience with high-performance, large-scale distributed systems.
Ability to scale research ideas into systems that run in production.
A proven track record of execution in ambiguous, fast-moving environments.
Excellent communication with technical and non-technical stakeholders.
Enjoys taking extreme ownership.
Passion for AI’s transformative potential.
Motivation to work on high-leverage infrastructure and systems supporting research across multiple tracks.

Nice To Haves

Experience with large-scale model training and inference.
Experience implementing LLM post-training recipes.
Deep familiarity with vLLM or SGLang, Ray, and Kubernetes.
Experience with distributed checkpointing and multi-node, multi-GPU training.
Experience with asynchronous training and inference frameworks or libraries such as VeRL, ROLL, SkyRL, AReaL, or CleanRL.

Responsibilities

Work backwards from the long-term research vision to identify and prioritize infrastructure and tooling work that connects research ambitions to near-term business value.
Rapidly prototype, build, and deploy AI systems for real users, real data, and production constraints.
Iterate based on user feedback and partner with stakeholders to ensure solutions are used, trusted, and adopted.
Embed within customer environments, build trust, and act as a technical partner on applied AI, identifying opportunities in customers’ data and infrastructure.
Contribute features, patterns, and abstractions learned from customer work to Mosaic, Percepta’s core product.
Partner with Embedded Product Managers to turn learnings into reusable playbooks for future customer transformations.