Top Tech Jobs & Startup Jobs

Own the end-to-end loop from simulation to physical robots: integrate and maintain physics simulators, build and operate teleoperation rigs, bring up and maintain robot platforms, and ensure reliable drivers, calibration pipelines, data recording, and deployment workflows for reproducible experiments.

Design, train, and evaluate learning algorithms connecting foundation models to physical systems for embodied agents. Focus on world models, action models, RL, and learning from demonstration in simulation and real-world deployments. Own end-to-end research projects, run large-scale experiments, collaborate with engineers and neuroscientists, and apply neuroscience principles to improve representations and behaviors.

Design, build, and maintain a production data platform for research: ingestion, ETL/ELT, object stores, SQL and NoSQL schema design, indexing, query optimization, observability, and high-throughput data movement. Collaborate with researchers, ML engineers, and infra to produce reliable interfaces for analytics, model training, and long-term archival while optimizing performance and maintainability.

Design and build an end-to-end, high-throughput dataloading stack for massive multimodal datasets: formatting, preprocessing, filtering, sharding, caching, and streaming data to distributed GPU training with observability, reliability, and performance benchmarking.

Build scalable agentic frameworks and reproducible experimental pipelines that integrate frontier multimodal models with rich neuroscience data. Implement orchestration, MLOps, and cloud infrastructure, balancing research and production engineering on a small collaborative team.

Develop and research multimodal foundation models focused on brain-scale data. Write production-quality code, build large-scale training and evaluation pipelines, establish benchmarks, and collaborate across computer vision, neuroscience, NLP, and robotics to shape foundational technical decisions.

Optimize training and inference performance of foundation models by writing and tuning CUDA/Triton GPU kernels, profiling and removing bottlenecks, implementing low-precision and mixed-precision strategies, optimizing MoE routing and expert dispatch, integrating high-performance libraries, and collaborating closely with researchers to translate model ideas into scalable, efficient implementations.

Develop, scale, and productionize state-of-the-art computer vision foundation models for multimodal data pipelines. Move research code to production, optimize inference and memory, build reliable ML workflows, and collaborate closely with researchers and engineers on large-scale AI research.

Design, build, and optimize distributed training systems for foundation models across thousands of GPUs. Implement advanced parallelism, fault-tolerance, profiling, and tooling to enable large-scale, reproducible ML experiments in cloud/HPC environments.