Staff Research Engineer – RRI Harness

Sequen · via Nocode Hub

💰 $300,000 - $350,000/yr 🌎 Remote ⌛ Posted 6 days ago

Build and evolve the Python multi-agent runtime, including orchestration, tool contracts, context and memory management, compaction, and model adapters across LLM providers
Design and run experiments on agent prompting, tool design, context and memory strategies, and model choice
Ship changes that measurably improve RRI results
Deliver a fixed evaluation set, scorecard with measured noise bands, per-change harness evaluations, and component ablations
Engineer resilience across GPU failures, node replacement, rolling upgrades, provider errors, and credit limits
Build event streams, cross-session analysis, per-agent cost and token accounting, and MCP administration tooling
Identify and reduce agent idling, looping, and excessive token usage
Design interfaces with the Go control plane and GPU proxy owners
Partner with applied scientists and forward-deployed engineers to diagnose and resolve production client issues
Own the core runtime of Sequen's autonomous research engine

7+ years of experience building production software
Strong Python skills
History of distributed or long-running systems
Experience building with LLMs in agentic loops, including tool calling, prompt and context management, streaming, and retries
Ability to debug across process, container, and network boundaries
Experience with Kubernetes and GPU nodes
Ability to read a training script, understand a ranking metric, and distinguish real regressions from noise
Applied research experience with ML or LLM experiments, baselines, and ablations
Experience with evaluation infrastructure, test harnesses, benchmarks, experiment tracking, or ML CI
Working knowledge of Go
Experience with on-prem, BYOC, or air-gapped environments and security constraints
Experience with PyTorch training at scale, GPU scheduling, or ML platforms
Contributions to agent frameworks, eval harnesses, or ML tooling are advantageous