Machine Learning Engineer
Computer Vision, NLP & Automatic Speech Recognition
I turn research into working AI systems — from court transcription pipelines to vision-based livestock tracking and generative image tools.
01 / About
Manchester, UK
Machine Learning Engineer specializing in computer vision, NLP, and automatic speech recognition (ASR), with strong experience in experimental evaluation, model benchmarking, and research-driven system development.
Published author with hands-on experience translating academic research into applied AI systems. Proficient in PyTorch, JAX, Transformers, and modern ML frameworks. Experienced in designing and evaluating AI pipelines for real-world applications — from court transcription systems to precision livestock monitoring and generative AI tools.
02 / Experience
Sep 2025 – Present
Leading research-driven development of AI-based ideation and image-editing tools. Integrated Segment Anything Model (SAM) for intelligent object segmentation and selection. Optimized model inference and interactive performance for real-time user workflows. Mentoring junior researchers on ML experimentation and engineering best practices.
Dec 2024 – May 2025
Designed and evaluated an AI-based automatic transcription system for criminal justice applications. Benchmarked multiple commercial and open-source ASR models using standardized error metrics. Developed a unified transcription pipeline aggregating best-performing models for improved accuracy. Authored detailed research documentation covering methodology, experiments, and deployment guidelines.
Aug 2023 – Jan 2024
Developed AI-powered tools for text-to-image, text-to-video, and text-to-3D generation. Integrated machine learning models into high-performance WebGL-based interactive systems.
Aug 2022 – Jan 2024
Built interactive 2D/3D product configurators and AI-enabled virtual try-on systems.
May 2019 – Jul 2022
Led end-to-end development of a scalable online design platform. Managed engineering teams and ensured system scalability and performance.
03 / Projects
Developed a full-stack transcription system using React, Flask, and WhisperX. Enabled transcription review, editing, subtitle embedding, and export workflows.
Built a collaborative online design and video editing platform. Integrated text-to-image and text-to-video generative AI tools with real-time collaboration.
Research-driven AI-based ideation and image-editing tool using Segment Anything Model (SAM) for intelligent object segmentation and selection.
Designed and evaluated an AI-based automatic transcription system for criminal justice applications. Benchmarked multiple ASR models for accuracy on court audio.
ViT-based multi-camera tracking and re-identification framework for precision livestock monitoring with MOTA ~98.7–99.3% and IDF1 >99%.
Interactive 2D/3D product configurators and AI-enabled virtual try-on systems built with WebGL for e-commerce applications.
04 / Publications
Evaluated modern ASR systems including Whisper, Deepgram, Azure, Google, IBM, and Amazon Transcribe. Conducted large-scale benchmarking using WER, CER, MER, WIL, and WIP on real-world court audio. Analyzed robustness across accents, noise conditions, and speaker variability.
Designed a ViT-based multi-camera tracking and re-identification framework for precision livestock monitoring. Achieved high performance with Multi-Object Tracking Accuracy (~98.7–99.3%) and IDF1 scores >99%. Addressed challenges including occlusion, viewpoint changes, and varying illumination.
05 / Skills
06 / FAQ
I specialize in computer vision, natural language processing (NLP), and automatic speech recognition (ASR). My work spans model benchmarking, experimental evaluation, and translating academic research into applied AI systems — including transcription pipelines, object tracking frameworks, and generative AI tools.
Yes, I am open to research collaborations, particularly in areas related to computer vision, NLP, ASR, and applied AI. I have experience co-authoring peer-reviewed publications and working on interdisciplinary research projects. Feel free to reach out to discuss potential collaborations.
I primarily work with Python, PyTorch, JAX, and Transformers for machine learning. For web development I use React, Next.js, Flask, and Django. I also have strong experience with WebGL, Three.js, and Fabric.js for interactive graphics, and AWS services (S3, EC2, SageMaker, Bedrock) along with Docker for cloud and MLOps.
I am open to consulting and freelance opportunities in AI/ML system design, model evaluation, and building end-to-end AI-powered applications. Whether you need help benchmarking models, building a transcription pipeline, or integrating generative AI tools, feel free to get in touch.
07 / Contact
Open to consulting, freelance projects, and full-time roles in ML engineering and applied research.
izeeshanafzal@outlook.com