Senior Computer Vision Engineer

altrocks tech pte. ltd.; Shenton Way; On-site; SGD 120,000 - 180,000; Full time

Job summary

Altrocks Tech Pte. Ltd. in Singapore seeks an experienced software engineer to architect and deploy large-scale video analytics pipelines across hundreds of feeds, using Python and C++ with DeepStream, GStreamer, or OpenCV.

You will translate model outputs into lightweight telemetry, interface with VMS/NVR systems, optimize latency, and collaborate on Kafka/RabbitMQ for streaming, Docker for deployment, and distributed queues.

Qualifications

• 5+ years of software engineering experience with 3+ years deploying large-scale video analytics or telemetry pipelines.
• Elite proficiency in Python and performance-critical C++.
• Deep expertise in PyTorch/TensorFlow, NVIDIA DeepStream SDK, GStreamer, Docker, and distributed message queues.
• Strong understanding of real-time streaming protocols and enterprise security transport configurations.

Responsibilities

• C2-Scale Analytics Pipelines: Architect robust Python/C++ pipelines using DeepStream, GStreamer, or OpenCV for hundreds of live feeds.
• Event Telemetry & Alert Generation: Translate model outputs into lightweight JSON/Protobuf payloads and stream to C2 brokers via Kafka/MQTT/gRPC.
• Multi-Camera Tracking & Fusion: Develop cross-camera Re-ID and spatial-temporal tracking across sites.
• VMS & Enterprise Integration: Build plugins/middleware for NVRs, VMS, and access control layers.
• Performance Tuning: Profile and scale inference architectures to maintain sub-second latency.
• Perimeter Security Sensors: Include tripwires and IoT occupancy sensors as part of the system.
• Hardware Integration Layers: Work with PLCs, Wiegand-to-IP, OSDP, and related hardware interfaces.

Skills

Python

C++

Video analytics

Distributed systems

Tools

NVIDIA DeepStream SDK

GStreamer

Docker

Kafka

RabbitMQ

PyTorch

TensorFlow

Job description

• C2-Scale Analytics Pipelines: Architect robust Python/C++ pipelines using frameworks like NVIDIA DeepStream, GStreamer, or OpenCV to execute concurrent video analytics across hundreds of live camera feeds.


• Event Telemetry & Alert Generation: Translate computer vision model outputs (bounding boxes, track IDs) into lightweight, structured JSON/Protobuf metadata payloads. Stream these alerts instantly to C2 event brokers using Kafka, MQTT, or gRPC


• Multi-Camera Tracking & Fusion: Develop algorithms for cross-camera re-identification (Re-ID) and spatial-temporal tracking, allowing the C2 system to map a subject’s trajectory across a physical site or facility layout.


• VMS & Enterprise Integration: Write custom plugins and middleware to interface directly with enterprise-grade Video Management Systems, network video recorders (NVRs), and unified access control physical layers.


• Performance Tuning for Command Centers: Profile, benchmark, and scale inference architectures using tools like TensorRT or ONNX Runtime to maintain strict sub-second end-to-end latency boundaries from camera sensor to C2 dashboard.


• Perimeter Security Sensors: Virtual/physical tripwire sensors, laser barriers, and IoT occupancy sensors.


• Hardware Integration Layers: Dry contact relays, Wiegand-to-IP converters, OSDP (Open Supervised Device Protocol), and Programmable Logic Controllers (PLCs).

Requirements:

• Experience: 5+ years of software engineering experience, with 3+ years specialized in deploying large-scale video analytics or telemetry pipelines integrated with operational monitoring systems (C2, VMS, or Smart City platforms).


• Programming Languages: Elite proficiency in Python and performance-critical C++.


• Frameworks & Infrastructure: Deep expertise in PyTorch/TensorFlow, NVIDIA DeepStream SDK, GStreamer, Docker, and distributed message queues (Kafka/RabbitMQ).


• Networking & Protocols: Thorough understanding of real-time streaming protocols (RTSP, WebRTC, SRT, HLS) and enterprise security transport configurations.