komal Sukheja

I’m a Machine Learning Engineer focused on turning data and computer vision ideas into production-ready ML models and data pipelines using Python. I’ve worked on real-time computer vision for security use cases, building systems designed for measurable performance gains.

At Cycomverse, I engineered feature extraction and preprocessing workflows with Pandas, NumPy, and visualization tooling to streamline dataset preparation for production ML training. I developed and deployed real-time person detection and object counting models using PyTorch with YOLO, and I also built anomaly detection solutions for suspicious activity and shoplifting. Through hyperparameter tuning and rigorous testing, I improved inference speed by 20% and increased prediction consistency.

Earlier, I strengthened my data and modeling foundation through roles as a Junior Consultant and Data Analyst Apprentice, including consolidating and cleaning multi-subsidiary financial data with Excel automation and ETL processes for 100% data accuracy, and creating a £1.5M predictive cost model with scenario analysis. I also built KPI dashboards with Excel and SQL, improving data visibility and insight accessibility by 30%. I bring this end-to-end mindset into my projects—like a Smart Document Scanner web app (OpenCV + Streamlit) and ML-powered analytics apps—so the work is both technically sound and genuinely useful.