Awais Raza

Manager AI/ML

The Bank of Punjab • Full-time

Dec 2025 - Present • 5 mos

Develop, and deploy AI/ML models for banking automation, fraud detection, risk management, and customer insights. Implement MLOps best practices, ensuring efficient model training, deployment, and monitoring. Work with data engineers to develop scalable data pipelines for machine learning models. Optimize AI models for performance, accuracy, and regulatory compliance. Developing and managing Data Lake solutions. Integrate AI/ML solutions into core banking systems and digital banking platforms. Stay updated on emerging AI trends, deep learning, NLP, and cloud-based ML frameworks. Ensure adherence to banking regulations, data privacy, and AI ethics policies. Collaborate with software engineers, DevOps, and cybersecurity teams to ensure secure model deployment. Provide technical mentorship to ML engineers and ensure continuous skill development within the team. Develop APIs, microservices, or model-serving solutions to integrate ML models into applications. Work with DevOps and MLOps teams to automate and monitor ML workflows. Implement CI/CD pipelines for machine learning models to ensure smooth deployment. Optimize inference speed and reduce computational costs for real-time ML applications.

Senior Software Enigineer AI/ML

PITB • Full-time

May 2025 - Oct 2025 • 5 mos

ChatBot: Developed an AI-powered multilingual chatbot leveraging Large Language Models (LLMs) to provide context-aware answers from uploaded documents in both English and Urdu, supporting interaction through text and speech. Implemented document retrieval with embeddings for accurate responses, integrated Text-to-Speech (TTS) and Speech-to-Text (STT) for real-time voice communication, and optimized the system for natural, human-like interaction. Tech stack included Python, LangChain, Hugging Face Transformers, Milvus DB, and Coqui TTS ANPR: Developed a high-performance Automatic Number Plate Recognition (ANPR) system by integrating a custom OCR engine (GlassRunner) with YOLO-based object detection models, enabling real-time license plate and vehicle recognition. Optimized video processing pipelines using ffmpegcv, achieving a 35% increase in throughput through dynamic Area of Interest (AOI) configurations.Deployed the complete solution using Docker Swarm for scalable distributed environments.

Software Engineer AI/ML

Punjab Information Technology Board • Full-time

Oct 2023 - Apr 2025 • 1 yr 6 mos

FD: Built a robust facial detection and recognition system capable of handling over 15 million facial records using YOLOv8 for detection and ResNet100 for recognition. Integrated the Milvus vector database for high-speed similarity search and deployed the system across Kubernetes clusters for high availability and modular microservice architecture. Integrations with SDKs: Integrated a wide range of biometric, medical, and authentication devices using C# SDKs, including SecuGen, Digital Persona, Green Bit Multiscan, HikVision, RealScan Palm Reader, and Suprema. Implemented both 1:1 (verification) and 1:N (identification) workflows using built-in SDK features, enabling real-time capture, backend matching, and responsive operations. Also developed socket-based integrations for real-time data transmission and LED-based dynamic display systems for field operations. CRO: Designed and developed a large-scale national criminal fingerprint database using Neurotechnology SDKs. Managed backend logic in C# and SQL Server for template storage and metadata management. Successfully enrolled over 37 million fingerprint records from all provinces of Pakistan and enabled ultra-fast 1:N matching for real-time suspect identification and investigation support. CRI: Developed the Crime Record Identification (CRI) application in C#, allowing inter provincial search and identification of criminals through live fingerprint scans. Enabled ntegration with provincial criminal databases and biometric devices like SecuGen and Green Bit to ensure cross-province visibility of criminal records.