PulseSan Francisco, CADevelops low-latency, high-throughput inference services for OCR and multimodal models, optimizing batching, kernels, and autoscaling while evaluating serving frameworks. Requires 3+ years in performance engineering or ML systems with strong Python and GPU experience.
Develop and integrate real-time sensor perception and fusion algorithms for autonomous vehicles across land, air, sea, and space domains. Requires MS/PhD or 5+ years experience with multi-modal sensors (EO/IR/radar), ML deployment, and DoD clearance eligibility.
Own the research-to-production pipeline at Deepgram, turning experimental speech ML models into reliable, scalable production services. Partner with researchers on robust workflows, automated release gates, inference optimization, and feedback loops across hybrid GPU infrastructure.
Backend Engineer building and optimizing Deepgram's core inference services for speech processing, including networking, audio transcoding, latency/memory optimization, and distributed compute orchestration. Requires 3+ years experience with Rust (or C/C++) and Python.
Research Engineers at Distyl build and productionize post-training techniques (fine-tuning, RLHF, reward models, evals) to improve reliability and behavior of compound AI systems for enterprise customers. Requires strong applied ML experimentation skills and ownership of real-world outcomes.
Research Engineers at Distyl build and productionize reliable agentic AI systems and compound architectures for enterprise workflows. They design agents, develop evaluation frameworks, run experiments on reasoning and failure modes, and integrate into customer environments.