Senior Applied AI Engineer (Multimodal Perception & Reasoning)

Key Responsibilities

• Build, fine-tune, and deploy production-grade multimodal models for safety and security applications, with a focus on visual and video perception, language-assisted and multimodal reasoning, and temporal understanding of real-world environments
• Own the full applied ML lifecycle, including data collection, labeling strategies, and dataset curation, model fine-tuning, evaluation, and iteration, and deployment, monitoring, and continuous improvement in production
• Drive model performance in real-world conditions, optimizing for high precision and recall, low false positives and false negatives, and robustness to noise, lighting changes, occlusion, and domain shift
• Optimize models for edge and cloud deployment, including quantization, pruning, and model compression, latency, throughput, and memory optimization, and hardware-aware tuning for GPUs and edge accelerators
• Build and maintain training and inference pipelines that support scalable experimentation and evaluation, reproducibility and model versioning, and reliable production deployment
• Collaborate closely with infrastructure and systems engineers to integrate models into real-time perception pipelines, balance accuracy, performance, and cost constraints, and diagnose and resolve production inference issues
• Use real-world deployment feedback and metrics to drive data and model improvements

Required Qualifications

• 8+ years of experience in applied machine learning or AI systems
• Strong hands-on experience with vision, video, or multimodal models
• Proven experience taking models into production, not just research prototypes
• Deep understanding of model optimization (quantization, pruning, performance tuning)
• Proficiency in Python and modern ML frameworks (e.g., PyTorch)
• Experience evaluating models using real-world metrics and constraints
• Ability to operate independently and own complex technical systems end to end

Preferred Qualifications

• Experience with multimodal or vision-language models (CLIP-like, BLIP-like, or custom)
• Experience deploying models to edge or resource-constrained environments
• Familiarity with inference optimization stacks (ONNX, TensorRT, CUDA)
• Experience working on physical-world perception systems (video, sensors, environments)
• Background in safety, security, robotics, or autonomous systems
• Experience mentoring senior engineers or providing technical leadership

What Success Looks Like

• Models ship reliably and improve measurable safety outcomes
• Precision and recall improve while inference cost and latency decrease
• Edge and cloud inference pipelines operate at production scale
• Data and model iteration loops accelerate over time
• AI perception becomes a durable competitive advantage for VOLT AI