Muavia Shakeel

AI/ML Engineer at Kohinoor Textile Mills Limited (KTML) (2025-08 – Present)

Architected and deployed an end-to-end Visual AI Surveillance Platform across 40 live RTSP cameras, supporting people counting, speed-violation detection, facial recognition, vehicle classification, and ANPR.
Engineered two concurrent DeepStream 7.1 inference pipelines on an RTX 5090 using YOLOv8, TensorRT FP16/INT8, and OpenCV, delivering approximately 120 FPS with sub-500 ms processing latency.
Built FAISS-based facial recognition for 10,000 employee embeddings and a resilient camera layer with database-driven configuration, auto-reconnect, freeze detection, and headless operation.
Developed the React and FastAPI dashboard for live views, alerts, counts, ANPR records, speed violations, face logs, analytics, filters, and reports; used by 3-5 Administrative Department users.
Integrated Oracle event logging and evidence snapshots, then containerized services with Docker and Nginx. The production system has been live at one mill site since Nov 2025.
Built and launched the Yarn ASM WhatsApp Bot in Mar 2026 for five sales users, enabling remote yarn booking, status tracking, stock checks, reports, and approvals through Meta WhatsApp Cloud API, n8n, FastAPI, and Oracle Database.
Removed office and laptop dependency for booking workflows while reducing manual reporting and errors and improving access to sales information.

Management Trainee Officer (MTO) at Kohinoor Textile Mills Limited (KTML) (2025-02 – 2025-08)

Built a production CNIC OCR and scanner application using Python, A8Capture.dll, Google Vision API, Urdu-to-English translation, automatic CNIC formatting, and Windows executable packaging.
Deployed the solution across KTM and KGM in Jul 2025; it processes approximately 100-150 CNICs per day in 1-3 seconds each with about 99% accuracy, reducing manual entry and improving speed and data quality.

Research Assistant - AI/ML Projects at Hazara University (2023-10 – 2025-01)

Supported supervisor-led research by implementing selected AI/ML topics, running experiments, preparing datasets, and building reproducible development prototypes.
Developed and optimized models for anomaly detection, predictive analysis, and classification, with selected experiments reaching 95%+ accuracy.
Automated data-preprocessing pipelines, reducing processing time by 50%, and deployed proof-of-concept services through Flask and FastAPI.
Explored GPT, LLaMA, and RAG workflows for document retrieval and chatbot development.

Python Developer Intern at Kohinoor Textile Mills Limited (2023-06 – 2023-08)

Developed Python automation scripts for production-data analysis, improving workflow efficiency by 30%.
Integrated SQL-based inventory tracking and reporting, contributed to Flask/FastAPI REST APIs, and built an operational analytics dashboard.

About