Principal Computer Vision Architect Vision AI Platform

Ranchhill Software Solutions; Hyderabad; On-site; INR 4,000,000 - 7,000,000; Full time

Job summary

NAVA Vision AI is seeking a hands-on Principal Computer Vision Architect to own architecture, deployment, scaling, and productization of production-grade Vision AI solutions across edge, cloud, and hybrid environments. The role requires hands-on experience deploying and scaling CV products in real-world settings, not just PoCs.

The architect will define how the Vision AI platform runs across customer sites, select edge servers/GPUs, determine camera capacity, optimize models and pipelines, and

Qualifications

• Hands-on architecting production-grade Vision AI systems across edge, cloud and hybrid environments.
• Experience deploying and scaling computer vision products in real-world environments.

Responsibilities

• Architect end-to-end Vision AI systems covering camera ingestion, video pipelines, inference, edge, cloud and multi-site deployment.
• Design and optimize multi-camera pipelines using NVIDIA DeepStream, GStreamer, CUDA, TensorRT and Triton.

Skills

Software engineering

Vision AI expertise

Mentoring

Tools

NVIDIA DeepStream

Triton Inference Server

TensorRT

CUDA

OpenCV

Python

RTSP/video analytics pipelines

Docker

Kubernetes/K3s

AWS

Edge AI deployments

YOLO

Job description

Principal Computer Vision Architect Vision AI Platform

Location: Hyderabad
Experience: 10+ Years
Employment Type: Full-Time

About the Role

NAVA Vision AI is looking for a hands-on Principal Computer Vision Architect to own the architecture, deployment, scaling, and productization of production-grade Vision AI solutions across edge, cloud, and hybrid environments.

This role is for someone who has personally deployed and scaled Computer Vision products in real-world environments, not just trained models or built PoCs.

The architect will define how NAVA’s Vision AI platform runs across customer sites, how edge servers and GPUs are selected, how many cameras each configuration supports, how models and pipelines are optimized, and how deployments are standardized for repeatable multi-site rollout.

Key Responsibilities

• Architect end-to-end Vision AI systems covering camera ingestion, video pipelines, inference, edge infrastructure, cloud, storage, APIs, monitoring, and multi-site deployment.
• Design and optimize multi-camera pipelines using NVIDIA DeepStream, GStreamer, CUDA, TensorRT, and Triton Inference Server.
• Architect containerized deployments using Docker, Kubernetes/K3s, Helm, NVIDIA Container Toolkit, and GPU-enabled orchestration.
• Design edge, cloud, and hybrid architectures, including production deployments on AWS.
• Select and benchmark appropriate edge servers, GPUs, CPU, RAM, storage, and network requirements based on:
• Use case
• Number of cameras
• Resolution and FPS
• Number and complexity of models
• Latency and performance requirements
• Build a repeatable camera-to-hardware sizing and benchmarking framework for different Vision AI workloads.
• Optimize models and inference pipelines using TensorRT, ONNX, FP16, INT8, quantization, pruning, compression, batching, and model optimization techniques.
• Design scalable architectures supporting deployment from a single site to tens or hundreds of customer locations.
• Establish standards for remote provisioning, configuration, model deployment, upgrades, rollback, monitoring, and diagnostics.
• Help productize NAVA Vision AI into a standardized edge deployment/appliance that can be installed, connected to customer cameras, configured, and brought into production with minimal custom engineering.
• Build for production reliability including RTSP reconnects, camera failures, network issues, service recovery, offline operation, health checks, observability, and automatic restart.
• Define best practices for model deployment, video processing, GPU utilization, infrastructure sizing, monitoring, and production operations.
• Review customer environments and make architecture decisions for camera topology, edge/cloud placement, infrastructure sizing, and deployment strategy.
• Guide and mentor Computer Vision engineers on model architecture, performance, inference optimization, and production deployment.

Creative Problem-Solving

We are looking for someone who does not assume every operational problem requires a more complex Computer Vision model.

The right candidate should be able to simplify problems by combining Vision AI with approaches such as:

• ArUco / AprilTag / fiducial markers
• QR codes or visual identifiers
• RFID / BLE / UWB
• IoT sensors
• PLC or machine telemetry
• WMS / TMS / MES data
• Other physical or software signals

The goal is always to identify the simplest, most reliable and scalable way to solve the operational problem.

Required Experience

• 8+ years of software, AI, platform, or systems engineering experience.
• Significant hands-on experience building and deploying production Computer Vision systems.
• Strong experience with:
• NVIDIA DeepStream
• Triton Inference Server
• TensorRT
• CUDA
• OpenCV
• Python
• RTSP/video analytics pipelines
• Docker
• Kubernetes / K3s
• AWS
• Edge AI deployments
• Experience deploying multiple Computer Vision solutions or products across edge and cloud environments.
• Experience processing multiple concurrent camera streams in production.
• Strong understanding of object detection, tracking, segmentation, OCR, pose estimation, video analytics, and multi-model pipelines.
• Experience with YOLO or similar modern detection architectures.
• Experience benchmarking GPU performance, throughput, latency, memory utilization, and camera capacity.
• Experience designing highly available, observable, remotely manageable production systems.

Strongly Preferred

• Experience deploying Vision AI in manufacturing, warehousing, logistics, industrial, transportation, retail, or physical security environments.
• Experience deploying the same platform across multiple customer sites.
• Experience supporting deployments with tens or hundreds of cameras.
• Experience with NVIDIA Jetson, IGX, RTX, L4, A-series, or similar GPU platforms.
• Experience with Prometheus, Grafana, OpenTelemetry, CloudWatch, or similar observability tools.
• Experience designing edge appliances or remotely managed edge software platforms.