Research Engineer: Scalable RL & Enterprise AI

Tessera Labs; New York (NY); On-site; USD 200,000 - 300,000; Full time

Job summary

Tessera Labs is building an AI-powered transformation engine for large enterprises. We are seeking a Research Engineer to design and implement scalable, production-grade training, evaluation, and inference machinery for long-horizon agents in enterprise environments.

You will own the post-training stack, memory/context systems, and datasets, collaborating with researchers to validate approaches that produce reliable, verifiable improvements at scale.

Qualifications

• Experience training, fine-tuning, or post-training language models with measurable results.
• Strong memory/context design for long-running agents and architectures.
• Excellent Python and PyTorch (or JAX) programming skills and debugging distributed training.
• Ability to design, run, and interpret rigorous experiments with empirical rigor.

Responsibilities

• Build and scale post-training stacks including SFT and RL for long-horizon tool use.
• Develop memory/context machinery for long-running agents across runs.
• Create knowledge graphs and ontologies from enterprise data for reasoning.
• Design data generation and curation pipelines with synthetic landscapes.
• Implement and run RL environments and offline evaluation harnesses.
• Lead end-to-end experiments from design to deployment and production.
• Optimize training and inference throughput and memory usage.

Skills

LM training

Memory systems

Python

PyTorch

Distributed training

Experiment design

GPU infrastructure

Technical writing

Education

Advanced CS degree

Tools

Git

Job description

Tessera Labs is building an AI-powered transformation engine for large enterprises. We are seeking a Research Engineer to design and implement scalable, production-grade training, evaluation, and inference machinery for long-horizon agents in enterprise environments.

You will own the post-training stack, memory/context systems, and datasets, collaborating with researchers to validate approaches that produce reliable, verifiable improvements at scale.