Artificial Intelligence Engineer

Punjab Safe Cities Authority (PSCA) is Pakistan’s leading smart city initiative focused on enhancing public safety and urban security across Punjab. Operating under the PSCA Act 2016, the Authority manages advanced Police Integrated Command, Control & Communication Centers (PPIC3) in major cities and is expanding to cover all districts. PSCA leverages AI-driven systems for traffic monitoring, road safety, and real-time surveillance of critical public areas, ensuring 24/7 citizen protection and security for high-profile events. The organization also champions citizen-centric solutions such as virtual police centers and mobile safety applications, making security services more accessible and inclusive. This role offers the opportunity to contribute to building safer, smarter, and more resilient urban environments.

Key Responsibilities

- Design, train, and optimize Computer Vision models for object detection, tracking, segmentation, face/person re-identification, license plate recognition, crowd analysis, and anomaly detection across extensive CCTV networks.
- Implement and fine-tune state-of-the-art object detection models including YOLO family (YOLOv8, YOLO11+), RT-DETR, DETR variants, Grounding DINO, and Segment Anything (SAM/SAM 2) for real-time surveillance and traffic management.
- Develop and deploy Vision-Language Model (VLM) solutions such as CLIP, LLaVA, Qwen-VL, and Florence-2 for video understanding, natural language video search, scene description, and automated incident summarization.
- Build LLM-powered applications for automated reporting, incident log analysis, emergency call triage, intelligent decision support, and citizen-facing assistants.
- Architect agentic AI workflows and Retrieval-Augmented Generation (RAG) pipelines using LangChain and LangGraph, integrating multiple tools, data sources, and models into reliable multi-step systems.
- Apply Transformer-based architectures across vision, language, and multimodal tasks, including fine-tuning techniques like LoRA/QLoRA and prompt engineering.
- Optimize AI models for real-time and edge deployment through quantization, pruning, and inference acceleration using frameworks such as TensorRT, ONNX Runtime, OpenVINO, and NVIDIA DeepStream.
- Collaborate with software engineering teams to integrate AI components into existing platforms via APIs and microservices, ensuring scalability, reliability, and low latency.
- Manage the full machine learning lifecycle: data collection, annotation, preprocessing, model training, evaluation, deployment, monitoring, and documentation.
- Work closely with law enforcement and operations personnel to translate real-world challenges into effective AI solutions.
- Stay updated on the latest AI research and evaluate emerging models and techniques for potential adoption.

Required Qualifications

- Bachelor’s or Master’s degree in Computer Science, Artificial Intelligence, Data Science, or a related technical field.
- Strong foundation in computer science principles including algorithms, data structures, and system design with practical application to AI.
- Proven hands-on experience in Computer Vision, including image/video analysis, object detection, multi-object tracking (e.g., ByteTrack, DeepSORT), and segmentation.
- Practical experience with modern object detection and segmentation models (YOLO family, DETR/RT-DETR, SAM) including training, fine-tuning, and deployment.
- Solid understanding of Transformer architectures and attention mechanisms, including Vision Transformers (ViT).
- Experience working with LLMs and VLMs, including fine-tuning, prompt engineering, and deploying open-source models such as Llama, Qwen, and Mistral.
- Hands-on experience with LangChain and LangGraph for building RAG systems, AI agents, and multi-step workflows, along with vector databases like FAISS, Milvus, Qdrant, or Chroma.
- Proficiency in Natural Language Processing (NLP) for applications such as automated alerts, speech-to-text, and incident log processing.
- Strong background in pattern recognition and neural networks for anomaly detection and predictive modeling.
- Excellent programming skills in Python, with working knowledge of C++ or Java.
- Familiarity with ML frameworks and libraries including PyTorch, TensorFlow, scikit-learn, Hugging Face Transformers, and OpenCV, as well as data tools like pandas, NumPy, and SQL.
- Experience with software development best practices including Git, REST APIs, microservices, Docker, and Linux environments.

Preferred Qualifications and Benefits

- Understanding of real-time systems, distributed architectures, and large-scale data stream processing technologies such as Kafka and Redis, especially in security or urban infrastructure contexts.
- Experience with model serving and MLOps tools like vLLM, Triton Inference Server, MLflow, and Kubernetes.
- Familiarity with edge AI hardware such as NVIDIA Jetson and GPU-accelerated video analytics pipelines.
- Contributions to publications, open-source projects, or a portfolio showcasing deployed AI solutions.
- Opportunity to work on some of Pakistan’s largest real-world AI deployments with direct impact on public safety.
- Access to large-scale, diverse datasets and advanced computing infrastructure.
- Be part of a mission-driven organization shaping the future of smart cities.
- Professional growth in an innovative environment that embraces cutting-edge technology.