Computer Vision Engineer | Video Analytics

Samriddhi Automations Pvt. Ltd.; Hinoba-an; On-site; PHP 900,000 - 1,500,000; Full time

Job summary

Samriddhi Automations Pvt. Ltd. in the Philippines seeks an AI Product Manager / Head of Software VA to own computer vision models end‑to‑end, from dataset curation and model training to real‑time inference and production deployment across multi‑camera sites.

You will collaborate with product, backend, deployment and QA teams to build robust inference pipelines using GStreamer/FFmpeg and modern AI toolchains, optimize for edge devices, and ensure seamless integration with NVR/VMS platforms.

Qualifications

• 3–6 years of hands‑on computer vision or deep learning experience, with production video analytics or surveillance focus
• B.Tech/M.Tech in Computer Science, Electronics or related field
• Strong Python and PyTorch or TensorFlow skills, with OpenCV and NumPy experience
• Comfortable with C++ for performance‑critical modules
• Experience with object detection, multi‑object tracking, OCR/ANPR and/or face recognition pipelines
• Hands‑on experience with ONNX, TensorRT, DeepStream/GStreamer, Docker and Linux
• Understanding of RTSP/ONVIF and video codecs like H.264/H.265

Responsibilities

• Design, train and fine‑tune CV models for surveillance use cases
• Build inference pipelines for live camera streams using GStreamer, FFmpeg or equivalent
• Optimize models for edge and GPU deployment using quantisation, pruning and toolchains
• Own annotation guidelines, dataset versioning, augmentation and retraining from field data
• Expose video analytics through APIs and metadata standards for NVR/VMS and dashboards

Skills

Python

PyTorch

TensorFlow

C++

OpenCV

NumPy

RTSP/ONVIF

Edge deployment

Education

B.Tech/M.Tech in Computer Science, Electronics or related field

Tools

GStreamer

FFmpeg

DeepStream

ONNX

TensorRT

OpenVINO

Docker

Linux

Job description

3–6 years

reporting_to

AI Product Manager / Head of Software VA

work_nature

AI / Technical / Production-focused

Where this role sits

Sparsh is building an AI-driven video analytics stack covering ANPR, face recognition, people and vehicle analytics, intrusion and perimeter detection.

This role sits within the Software & Video Analytics (VA) team and works closely with product, backend, deployment and QA teams.

You will own computer vision models end‑to‑end, from dataset curation and model training to real-time inference and production deployment across large multi-camera sites.

The role

This role is:

What you’ll work on

Design, train and fine-tune detection, tracking, classification and recognition models for real-world surveillance use cases

Build inference pipelines for live RTSP/ONVIF camera streams using GStreamer, FFmpeg, DeepStream or equivalent technologies

Optimise models for edge and GPU deployment using quantisation, pruning, ONNX, TensorRT and OpenVINO/NPU toolchains

Own annotation guidelines, dataset versioning, augmentation, hard-negative mining and periodic retraining from field data

Production Integration

Expose video analytics through APIs, event schemas and metadata standards for NVR/VMS, dashboards and third‑party platforms

Who this role is for

• 3–6 years of hands‑on computer vision or deep learning experience, with at least 2 years in production video analytics or surveillance
• B.Tech/M.Tech in Computer Science, Electronics or a related field
• Strong Python and solid PyTorch or TensorFlow skills, with experience in OpenCV and NumPy
• Comfortable with C++ for performance‑critical modules
• Experience with object detection, multi‑object tracking, OCR/ANPR and/or face recognition pipelines
• Hands‑on experience with ONNX, TensorRT, DeepStream/GStreamer, Docker and Linux
• Understanding of RTSP/ONVIF, H.264/H.265 video codecs and camera imaging fundamentals

What you’ll learn here

• How production‑grade computer vision models are built and deployed on real camera streams
• How AI models are optimised for edge devices and GPU platforms
• How video analytics systems are benchmarked across lighting, weather, camera angles and Indian scene conditions
• How computer vision models integrate with NVR/VMS platforms and customer systems
• How AI products move from proof of concept to production across large multi-camera sites
• Work on India's first STQC-certified, Made-in-India surveillance platform deployed at national scale
• Ship AI that runs in the real world on live camera streams and in the field
• Gain full‑stack exposure across hardware, firmware, software and AI under one roof

Ready to build AI that works in the real world?