| name | ml-systems-engineer-rl-engineering |
| description | Guides ML systems engineering for reinforcement learning—distributed training platforms, rollout
workers and vectorized environments, replay buffers, policy/critic serving for train loops,
checkpointing and experiment tracking, sim-to-real hooks, and RL training reliability.
Use when building RL training infrastructure, scaling PPO/SAC-style jobs, debugging unstable
distributed rollouts, designing env APIs, or exporting policies for inference—not for supervised
ML product modeling (data-scientist), LLM RAG/agents (ai-engineer), safeguard classifiers
(ml-research-engineer-safeguards, ml-infrastructure-engineer-safeguards), general GPU serving
without RL (ml-infrastructure-engineer-safeguards), or app latency profiling (performance-engineer).
|
Machine Learning Systems Engineer, RL Engineering
When to Use
- Design RL training platform — controllers, workers, resource scheduling
- Implement rollout collection — vectorized envs, async actors, trajectory buffers
- Operate distributed training — data parallel, parameter servers, gradient sync patterns
- Manage replay buffers — prioritization, storage, sampling at scale
- Wire checkpointing — policy/value nets, optimizer state, resume after preemption
- Integrate experiment tracking — seeds, configs, metric schemas, artifact lineage
- Connect simulators — Gymnasium-style APIs, custom env servers, batch stepping
- Export policies for batch eval or downstream inference path
- Debug training instability — NaNs, reward scale, worker desync, straggler GPUs
- Plan GPU/memory layout for actor vs learner processes
When NOT to Use
- Churn models, A/B tests, classical supervised pipelines →
data-scientist
- Production LLM features, agents, RAG →
ai-engineer
- Safeguard/moderation inference gateways →
ml-infrastructure-engineer-safeguards
- Safety classifier research →
ml-research-engineer-safeguards
- CI/CD and generic K8s ops →
devops, cluster-deployment-engineer
- DC-wide GPU supply programs →
data-center-compute-supply-efficiency
- HTTP API p99 without RL training context →
performance-engineer
- RL algorithm theory only (no systems) →
ai-researcher for literature; stay systems-focused here
Related skills
| Need | Skill |
|---|
| Supervised ML and statistical eval | data-scientist |
| General AI research methodology | ai-researcher |
| Inference gateways and model serving | ml-infrastructure-engineer-safeguards |
| Training cluster / K8s jobs | cluster-deployment-engineer |
| Pipelines and GitOps | devops |
| GPU capacity at facility level | data-center-compute-supply-efficiency |
| Serving latency and load tests | performance-engineer |
| Product agents using RL outcomes | ai-engineer |
Core Workflows
1. RL systems framing
Env contract, on/off-policy, scale targets.
See references/rl_systems_framing.md.
2. Training platform architecture
Controllers, workers, scheduling.
See references/training_platform_architecture.md.
3. Environments and rollouts
Vectorization, trajectory format.
See references/environments_rollouts.md.
4. Replay, checkpoints, experiments
Buffers, resume, tracking.
See references/replay_checkpoints_experiments.md.
5. Evaluation and policy export
Eval harness, deployment handoff.
See references/evaluation_policy_export.md.
6. Reliability and observability
Stability, metrics, incident debug.
See references/reliability_observability_rl.md.
Outputs
- Architecture doc — actor/learner topology, data flow, failure domains
- Env API spec — observation, action, reward, reset, seed semantics
- Runbook — launch, resume, preempted job recovery, scale-out
- Config template — hyperparameters + infra knobs versioned together
- Metric dashboard spec — reward, length, KL, GPU, steps/sec, queue depth
- Policy export package — weights, normalization stats, eval report
Principles
- Reproducibility — seed envs, log config hash, pin sim versions
- Separate rollout from learn — scale collectors and learners independently
- Deterministic resume — checkpoint includes optimizer and buffer cursor when needed
- Observe the MDP — log reward components, not only scalar return
- Fail fast on desync — version mismatch between workers is a top incident class
