Mirendil
Open positions
Member of Technical Staff, Distributed Systems Engineer
Mirendil Mirendil is a tech-first company focused on solving core bottlenecks that unlock step-change acceleration across science and technology. Our first goal is to democratize frontier AI R&D across scientific…
Member of Technical Staff, Model Evaluation
Research engineer building evaluation infrastructure for a frontier AI lab that trains its own models end-to-end: designing eval frameworks, automated eval pipelines and regression detection, agent-assisted inspection workflows, and observability tooling for training runs, in partnership with post-training and RL teams.
Member of Technical Staff, Pretraining
Engineer on a frontier AI research company's pretraining stack, working across data processing, model architecture, distributed training infrastructure, and low-level optimization to train large models end-to-end at scale.
Member of Technical Staff, Post-Training, RL Environments
Research engineer building the data pipelines and sandboxed execution environments that power reinforcement-learning post-training for a frontier AI lab. The role owns data collection for long-horizon RL tasks, reward-hacking prevention, and scalable multi-agent execution environments.
Member of Technical Staff, Kernels
Mirendil is hiring an engineer to design, implement, and optimize custom ML kernels for its frontier AI model training stack, working at the hardware/software intersection to improve performance, reduce memory bandwidth bottlenecks, and enable low-precision arithmetic formats.
Member of Technical Staff, Infrastructure
Infrastructure engineer at a frontier AI research lab, owning the compute and cloud foundation for model training and inference: Kubernetes/GPU cluster operations, sandboxed secure execution for agents, long-running agent state management, and cross-cloud networking.
Member of Technical Staff, Inference
Owns the inference stack for a frontier AI research company, building high-throughput, low-latency serving systems for large models. Works across GPU/accelerator optimization, distributed inference frameworks (vLLM, SGLang, TensorRT-LLM), and inference-time optimizations like speculative decoding, quantization, and KV cache management.