Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

AI/ML Engineer with 2+ years of experience specializing in NLP and ASR systems, currently completing an MSc in AI and Adaptive Systems at the University of Sussex. I design scalable production pipelines, fine-tune transformer models (BERT and Whisper), and deploy containerized microservices for real-time performance. I bridge experimental research with high-performance engineering to deliver robust, production-ready AI solutions—improving accuracy by 18% and reducing latency by 25% through careful optimization, batching, and reproducible ML workflows.…AI/ML Engineer with 2+ years of experience specializing in NLP and ASR systems, currently completing an MSc in AI and Adaptive Systems at the University of Sussex. I design scalable production pipelines, fine-tune transformer models (BERT and Whisper), and deploy containerized microservices for real-time performance. I bridge experimental research with high-performance engineering to deliver robust, production-ready AI solutions—improving accuracy by 18% and reducing latency by 25% through careful optimization, batching, and reproducible ML workflows.

Muhammad Umar Moeen

AI Engineer, Data Scientist, AI Developer, +2





AI/ML Engineer with 2+ years of experience specializing in NLP and ASR systems, currently completing an MSc in AI and Adaptive Systems at the University of Sussex. I design scalable production pipelines, fine-tune transformer models (BERT and Whisper), and deploy containerized microservices for real-time performance. I bridge experimental research with high-performance engineering to deliver robust, production-ready AI solutions—improving accuracy by 18% and reducing latency by 25% through careful optimization, batching, and reproducible ML workflows.…AI/ML Engineer with 2+ years of experience specializing in NLP and ASR systems, currently completing an MSc in AI and Adaptive Systems at the University of Sussex. I design scalable production pipelines, fine-tune transformer models (BERT and Whisper), and deploy containerized microservices for real-time performance. I bridge experimental research with high-performance engineering to deliver robust, production-ready AI solutions—improving accuracy by 18% and reducing latency by 25% through careful optimization, batching, and reproducible ML workflows.

Available to hire

AI/ML Engineer with 2+ years of experience specializing in NLP and ASR systems, currently completing an MSc in AI and Adaptive Systems at the University of Sussex. I design scalable production pipelines, fine-tune transformer models (BERT and Whisper), and deploy containerized microservices for real-time performance.

I bridge experimental research with high-performance engineering to deliver robust, production-ready AI solutions—improving accuracy by 18% and reducing latency by 25% through careful optimization, batching, and reproducible ML workflows.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Intermediate

Beginner

Language

English

Fluent

Urdu

Fluent

German

Beginner

Work Experience

Research Engineer - Deep Learning & NLP at FAST NUCES Lahore, Pakistan

January 1, 2024 - January 1, 2025

Researched transformer-based ASR architectures using LibriSpeech and domain-specific datasets for benchmarking. Fine-tuned Whisper, Wav2Vec2, and HuBERT models, reducing WER by 18% compared to baseline CNN-LSTM models. Implemented time and frequency masking improving robustness on noisy speech by 12%. Developed CNN-LSTM emotion recognition models using MFCCs achieving 85% classification accuracy. Performed systematic hyperparameter tuning to improve experimental reproducibility and model stability.

Machine Learning Engineer at Programmers Force Lahore, Pakistan

January 1, 2023 - January 1, 2024

Engineered a production NLP pipeline for adverse media detection processing 100k+ documents using transformer-based classification. Fine-tuned BERT/RoBERTa models achieving 92% accuracy and improving throughput by 40% via batch optimization. Optimized KYC models through tuning and pruning, yielding a 15% accuracy gain and 25% latency reduction. Deployed containerized ML services using FastAPI and Docker, achieving sub-250ms real-time response times. Streamlined tracking pipelines to reduce model iteration cycles by 30% and improve team reproducibility.