We are looking for a Vision AI Architect to lead the design and implementation of large-scale Vision AI platforms and solutions. The role requires deep expertise across the entire video and vision lifecycle: capture, encoding, transcoding, transport, preprocessing, model development, training, deployment, scaling, application integration, and production operations. You will define reference architectures, guide engineering teams, and lead customer engagements from solution discovery through deployment.

Programming Languages & Technology Stack

· Expert-level Python, C/C++ or similar core programming languages

· Robust experience with PyTorch, TensorFlow, OpenCV, CUDA, NVIDIA ecosystem, and AI acceleration frameworks.

· Understanding of microservices, REST APIs, containerization, and distributed systems.

Requirements

Video Systems, Media Processing & Streaming Architecture

· Deep expertise in video capture, ingestion, encoding, transcoding, packaging, streaming, storage, and distribution.

· Strong understanding of video codecs and standards: H.264, H.265/HEVC, AV1, MPEG-TS,



MP4, RTSP, RTP, WebRTC, SRT, HLS, DASH.

· Hands-on experience designing scalable video pipelines using FFmpeg, GStreamer, NVIDIA DeepStream, or equivalent frameworks.

· Experience with edge-to-cloud video transportation architectures and low-latency streaming systems.

· Knowledge of video quality optimization, bitrate adaptation, frame extraction, synchronization, and metadata management.

Experience handling large-scale video workloads across distributed environments.

Model Development, Training & Optimization

· Experience building and training production-grade vision models on custom datasets.

· Deep understanding of dataset design, annotation strategies, augmentation, transfer learning, and active learning.

· Model optimization using TensorRT, ONNX Runtime, quantization, pruning, and GPU acceleration.

· Experience deploying models across edge, cloud, and hybrid environments.

📌 Python Vision AI Architect (Thane)
🏢 PEOPLEOCITY
📍 Thane