Open positions
Member of Technical Staff - Engineering Lead, Compute Platform
Engineering Lead for Reflection's Compute Platform in London: you'd build and mentor a team of ~10 systems engineers running a Kubernetes-based multi-cloud compute fleet for large AI training runs. The role blends front-line leadership, hands-on systems work (multi-cloud scheduling, node health, GPU fleet), vendor/deal management, and co-designing fault tolerance with training teams.
Research Program Manager - Research Infrastructure
Research Program Manager at Reflection, an AI research lab building open frontier models. Day to day you drive cross-functional programs for training infrastructure and cluster reliability across pre-, mid- and post-training, triage incidents, and keep GPU-scale training runs (e.g. Megatron-based) healthy and unblocked.
Research Program Manager - Model Data & Evaluations Flywheel
A senior program manager role at Reflection's London office to build the model evaluations and AI safety function from scratch — defining eval frameworks, safety review processes, and external safety partnerships that plug into frontier open-model development. Requires 7+ years in technical program management, research ops, or ML engineering.
Research Program Manager - Model Development
A research program manager embedded with Reflection's pre-training ML and Data teams, coordinating frontier LLM development — experiment planning, data pipelines, run scheduling, and cross-team handoffs. Day to day, they build lightweight processes and drive decisions so researchers can move faster on open model development.
Member of Technical Staff - Pre-Training Infra
Builds and scales the distributed training infrastructure behind Reflection's frontier open-model pre-training: designing large training runs with researchers, optimizing GPU utilization, communication, and memory across thousands of GPUs, and debugging bottlenecks in the distributed training stack using frameworks like Megatron/DeepSpeed and NCCL.
Member of Technical Staff - Compute Platform
Builds and maintains Reflection's Kubernetes-based compute platform for AI model training across multiple cloud providers — handling cluster management, topology-aware GPU scheduling, node health, monitoring, and petabyte-scale storage. Requires deep systems engineering and GPU infrastructure experience (e.g., NCCL).
Member of Technical Staff - Platform Foundations
Platform/infrastructure engineer at Reflection, an open-weight AI research lab, building the company-wide foundations platform in London: golden paths for cloud and networking, security guardrails, CI/CD and build systems, observability, and cloud cost optimization using GCP/AWS, Kubernetes, Terraform/Pulumi, and Bazel.
Member of Technical Staff - Shared Services
Builds and operates the shared-services foundations platform powering Reflection's open AI model research and production environments — developer infrastructure, SRE reliability, and data/compute tooling. Day-to-day work spans Kubernetes container platforms, distributed systems, and services in Go, Rust, C++, and gRPC.
Member of Technical Staff - Applications
Full-stack engineer at AI research lab Reflection, building internal and customer-facing product surfaces (dashboards, admin tools, self-serve workflows) on top of the company platform. Day-to-day spans React/TypeScript UIs, FastAPI/Go APIs, OIDC auth and RBAC, event-driven integrations, and observability tooling.
Member of Technical Staff - Data Ingestion Engineer
Data Ingestion Engineer at Reflection, an AI research lab building open frontier models, builds and operates large-scale data ingestion systems (web crawling, extraction, normalization, versioning, delivery) that turn the open web into training corpora for pre-training pipelines, using tools like Ray, Beam, and Spark.
Member of Technical Staff - Safety
AI Safety engineer at Reflection's London lab who owns red-teaming and adversarial evaluation of the company's open LLMs, builds automated safety benchmarks, researches jailbreaks and defenses, and acts as gatekeeper ensuring model releases meet risk thresholds. Core tech: LLM safety, adversarial eval pipelines, ML systems, RLHF/RLAIF.
Member of Technical Staff - Data Quality Engineer (Pre-training)
Data Quality Engineer on Reflection's pre-training data team at an AI research lab building open foundation models. Day to day: define measurable data-quality standards, build automated QA pipelines, and monitor training data quality for LLM pre-training, primarily using Python and ML/LLM-based evaluation methods.
Member of Technical Staff - Data Quality Engineer (Post-training)
At Reflection, an AI research lab building open models, this London-based engineer owns upstream data quality for LLM post-training: analyzing expert datasets, turning 'good data' into measurable standards, and building automated QA pipelines (including LLM-as-a-Judge) that feed model training and evaluation. Core stack: Python, ML/LLM workflows, and large-scale data systems.
Member of Technical Staff - Evaluations
At Reflection, an AI research lab building open foundational models, you'll design and run LLM evaluations — building benchmarks, human preference and agentic evals, and feedback loops that connect data, evals, and model behavior to drive model improvements across pre-training, post-training, and applied teams.
Member of Technical Staff - Pre-Training
Research engineer role on the pre-training team at Reflection, an AI lab building open foundation models. Day to day you run experiments on scaling laws, data processing, optimizers and model architecture, implement SOTA deep learning methods, and optimize large-scale distributed training infrastructure using Python and PyTorch.
Member of Technical Staff - Post-Training
AI research lab Reflection is hiring a Member of Technical Staff (Post-Training) in London to turn pre-trained LLMs into aligned, general agents — building data generation pipelines, reward models, RL algorithms, and inference-time scaling. Requires strong ML fundamentals, large-scale LLM training experience, and solid engineering skills.
Member of Technical Staff - Engineering Lead, Data Ingestion
Engineering lead for Reflection's data ingestion team in London, building the web-scale crawling, extraction pipelines, and data lakes that turn the open web and licensed sources into training corpora for the company's open frontier AI models. A hands-on management role spanning Ray/Spark distributed compute, Airflow/Prefect orchestration, and large-scale dataset versioning and delivery.
Member of Technical Staff - Data Platform
Builds and operates the core data platform at Reflection, an AI research lab creating open-weight models — designing batch/streaming ingestion, orchestration, and storage/compute foundations using tools like Spark, Flink, Airflow/Dagster, Kafka, and lakehouse formats such as Iceberg and Delta Lake to power research, training, and production.
Member of Technical Staff, Data Flywheel
Hands-on research-engineering role on Reflection's Data Flywheel team in London: turning real-world model usage into evaluations, curated human/synthetic training data, and feedback loops that measurably improve LLMs. Day-to-day spans evaluation design, data pipelines and infrastructure, partner/data-source integration, and post-training reward signals.
Member of Technical Staff - Pre-Training Evals Infra
Builds evaluation infrastructure for an open-foundational-model AI research lab: designing eval APIs and data models, running distributed evaluation workloads across pre-training checkpoints on GPUs, profiling inference performance, and creating researcher-facing dashboards. Core stack: Python, C++, or Rust plus distributed systems.
Lightspeed