S
Stack AV Tech Stack
AI and autonomous systems for transportation safety and efficiency
Stack AV builds autonomous vehicle software with a heavy ML/systems engineering footprint: PyTorch, CUDA, TensorRT, Ray, and Apache Spark for training and inference, paired with real-time control layers (C++, FreeRTOS, gRPC). The stack and project list—long-range detection, trajectory generation, model deployment optimization, semantic search over multimodal data—reflect a company solving the hard problem of perception-to-control in self-driving. Hiring skews senior and staff-level engineering with ops support, suggesting they're scaling model iteration velocity and data pipeline reliability rather than early-stage exploration.
Tech Stack 32 technologies
Core StackPython C++ Apache Spark Iceberg PyTorch gRPC Go PostgreSQL Redis Kafka Docker Kubernetes Ray CUDA Triton TensorRT ONNX LinkedIn Recruiter Bazel Boost POSIX FreeRTOS SafeRTOS WebSockets QUIC WebRTC Tilt
What Stack AV Is Building
◆Challenges
- Development bottlenecks
- Low latency/high throughput processing
- Accelerate model iteration
- Handle long tail challenges
- Meeting safety and performance standards
- Understanding system performance
- Identifying suboptimal behavior
- Existing ml infra performance issues
- Scaling petabytes data processing
- Fault tolerant data pipelines
▲Active Projects
- Model deployment optimization
- Long-range detection capabilities
- Proprietary labeling and mapping tools
- Trajectory generation and controls features
- Real-time trajectory generation
- Triage event annotation
- Performance reporting on self-driving system
- Inference service using llms and vector databases
- Scalable semantic search service for multimodal data
- Infrastructure for ml training and inference workloads
Hiring Activity
15 roles · 8 in 30d
Department
Seniority
Company intelligence
