Software Engineer

At Mantari Industries, we're replacing yesterday's public sector weather systems with a full stack, AI-native platform that senses, models, and predicts Earth's atmosphere and oceans in near real time. Our mission is to sense Earth in its entirety - oceans, skies, and space - and close the gap between what nature does and what humanity knows. By building a distributed network of sensors fused with breakthrough AI, Mantari transforms raw environmental signals into actionable intelligence so that the world's most critical outcomes, from global trade and securing grids to defending nations, are guided not by uncertainty, but by foresight. Our autonomous edge sensing fleets, deep learning forecast models, and integrated customer decision engines will create the most precise weather intelligence on the planet-empowering industry, strengthening climate resilience, providing vital intelligence to American warfighters, and opening billion-dollar markets that don't exist yet.


Backed by premier venture capital and led by engineers and scientists who've landed rockets and pioneered novel AI forecast techniques, we move fast, own the hard problems end to end, and measure success in real world impact. If you want to build frontier tech with outsized consequences, welcome aboard.

Description

Own the guts of our modeling engine, everything between raw sensor data and a lightning-fast forecast API:

• Build and optimize CUDA kernels, JAX/PyTorch ops, and distributed training loops for physics-aware transformer models
• Stand up reproducible, containerized pipelines (Ray, Kubernetes, Slurm, or similar) that scale from a single GPU to multi-region GPU clusters in the cloud
• Automate data ingest and feature engineering from our growing sensor network (satellite, buoy, radar, etc.)
• Translate novel research prototypes into production-grade code

You'll work closely with Mantari's Head of Weather and AI modeling teams to operationalize next-generation forecast systems, bringing meteorological science and modern ML together at planetary scale.
Requirements

• Degree in Atmospheric Science, Computer Science, or related engineering / physical sciences field
• 5+ years writing high-performance Python/C++/Rust for GPU workloads
• Real-world experience designing and building cloud-native applications and infrastructure (GCP, AWS, or Azure)
• Deep fluency with at least one modern deep learning framework (JAX, PyTorch, TensorFlow, etc.)
• Proven track record shipping distributed training or inference systems at scale (100+ GPUs)
• Comfortable profiling kernels, diagnosing memory bottlenecks, and reading compiler logs when needed
• Strong grasp of modern DevOps: code reviews, CI/CD pipelines, observability, and containerization

Bonus Qualifications

• Experience with numerical weather prediction, remote-sensing data, or geospatial intelligence
• Background in physics-informed ML, simulation modeling, or data assimilation
• Contributions to open-source ML or scientific computing frameworks

How We Work

• Small, senior team; high autonomy, high expectations.
• Collaborate closely with the founding team and SMEs to push the boundaries of observation-driven AI-native weather forecasting, and turn weather into an advantage, not a roadblock.
• LA HQ required.
• Equity upside commensurate with being one of the first core team members. Competitive salary, full benefits, and hardware budget.

Weather shapes every decision made on Earth. At Mantari you'll write code that moves metal, launch sensors that tame storms, and see your work ship in weeks-not years. Join us to build the world's first private full-stack weather platform and leave your fingerprints on infrastructure that will outlast us all. Apply now-let's create the future of weather, together.