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

I am a Machine Learning Engineer with over four years of experience focused on computer vision and 3D reconstruction, delivering robust and production-ready AI models optimized for resource-constrained environments. I specialize in PyTorch, model quantization, and deploying deep learning models on embedded devices, consistently achieving high accuracy and improved inference performance. Throughout my career, I have contributed to multiple innovative projects including real-time driver monitoring, human mesh generation, and synthetic data pipelines. I excel at optimizing workflows, automating data processing, and integrating state-of-the-art methods to drive impactful AI solutions in both production systems and research environments.…I am a Machine Learning Engineer with over four years of experience focused on computer vision and 3D reconstruction, delivering robust and production-ready AI models optimized for resource-constrained environments. I specialize in PyTorch, model quantization, and deploying deep learning models on embedded devices, consistently achieving high accuracy and improved inference performance. Throughout my career, I have contributed to multiple innovative projects including real-time driver monitoring, human mesh generation, and synthetic data pipelines. I excel at optimizing workflows, automating data processing, and integrating state-of-the-art methods to drive impactful AI solutions in both production systems and research environments.

Muhammad Junaid Ahmad





I am a Machine Learning Engineer with over four years of experience focused on computer vision and 3D reconstruction, delivering robust and production-ready AI models optimized for resource-constrained environments. I specialize in PyTorch, model quantization, and deploying deep learning models on embedded devices, consistently achieving high accuracy and improved inference performance. Throughout my career, I have contributed to multiple innovative projects including real-time driver monitoring, human mesh generation, and synthetic data pipelines. I excel at optimizing workflows, automating data processing, and integrating state-of-the-art methods to drive impactful AI solutions in both production systems and research environments.…I am a Machine Learning Engineer with over four years of experience focused on computer vision and 3D reconstruction, delivering robust and production-ready AI models optimized for resource-constrained environments. I specialize in PyTorch, model quantization, and deploying deep learning models on embedded devices, consistently achieving high accuracy and improved inference performance. Throughout my career, I have contributed to multiple innovative projects including real-time driver monitoring, human mesh generation, and synthetic data pipelines. I excel at optimizing workflows, automating data processing, and integrating state-of-the-art methods to drive impactful AI solutions in both production systems and research environments.

Available to hire

I am a Machine Learning Engineer with over four years of experience focused on computer vision and 3D reconstruction, delivering robust and production-ready AI models optimized for resource-constrained environments. I specialize in PyTorch, model quantization, and deploying deep learning models on embedded devices, consistently achieving high accuracy and improved inference performance.

Throughout my career, I have contributed to multiple innovative projects including real-time driver monitoring, human mesh generation, and synthetic data pipelines. I excel at optimizing workflows, automating data processing, and integrating state-of-the-art methods to drive impactful AI solutions in both production systems and research environments.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Intermediate

Work Experience

Machine Learning Engineer at Rooya Vision

November 1, 2024 - Present

Revamped on-device algorithms for harsh acceleration and braking detection using accelerometer and GPS data, suppressing 90% false events. Delivered a speed violation detection and tailgating multi-task model achieving over 85% accuracy in real-world deployment. Optimized a custom multi-head model via pruning to achieve 2 FPS inference on Qualcomm Snapdragon SDM450, balancing accuracy and performance under hardware constraints.

Machine Learning Engineer at MTailor

May 1, 2024 - August 27, 2025

Led development of human mesh generation system using multi-view PIFu with ColMap poses, improving fit accuracy by 28%. Engineered synthetic data pipeline generating 200K+ SMPLX meshes via BlenderPython for model training. Built data processing pipeline using SAM and DepthAnything automating segmentation and anonymization. Optimized training workflows by integrating SkyPilot and ClearML reducing experiment setup time by 45%. Conducted A/B testing on fit algorithm improvements, resulting in 15% reduction in size-related returns.

Computer Vision Engineer at Motive (formerly KeepTruckin)

November 1, 2024 - August 27, 2025

Developed production-ready driver monitoring features detecting smoking and distraction behaviors with 94% accuracy. Architected multi-task YOLOx models with custom heads enforcing shared information between tasks. Reduced model size and inference time by 50% through 8-bit quantization using SNPE on Qualcomm devices. Delivered semantic search system using OpenAI embeddings and Chroma vector store supporting real-time accurate answers across internal documents. Decreased data preprocessing pipeline latency by 60% through parallel processing optimization. Created generic dataset package and automated labeling system using SAM, increasing annotation efficiency by 3x.

Software Engineer at Confiz Limited

February 1, 2022 - August 27, 2025

Refined end-to-end sales forecasting pipeline in PySpark using ensemble models and time-series analysis with 18% higher accuracy. Trained on enterprise ML pipelines using Scala2, AirFlow and Google Cloud Platform.

Junior Embedded Software Engineer at Vadion Pvt Limited

July 1, 2021 - August 27, 2025

Automated product deployment process using kickstart scripts, reducing installation time by 70%. Configured proxy servers using Nexus and Nginx to optimize network performance in production environments. Established CI/CD pipeline on TeamCity, cutting release cycle time from days to hours.

Research Intern, VisPro Lab at Information Technology University

March 1, 2021 - August 27, 2025

Collaborated with ETRI R&D on novel 3D reconstruction pipeline integrating deep learning with classical methods. Replaced SIFT and RANSAC with SuperPoint and SuperGlue for robust keypoint detection and matching, improving reconstruction quality in low-texture environments by 35%.

Embedded Software Engineer Intern at Mentor - A Siemens Business

September 1, 2019 - August 27, 2025

Upgraded diagnostic tool for testing LIN and CAN-FD bus protocols in automotive applications. Redesigned transport layer for CAN-FD protocol supporting 8-Mbits/sec in Audi-Porsche mainboard.

Machine Learning Engineer at Rooya Vision

November 1, 2024 - Present

Machine Learning Engineer at MTailor

May 1, 2024 - August 27, 2025

Led development of human mesh generation system using multi-view PIFu with ColMap poses, improving fit accuracy by 28%. Engineered synthetic data pipeline generating 200K+ SMPLX meshes via BlenderPython for model training. Built data processing pipeline using SAM and DepthAnything, automating segmentation and anonymization. Optimized training workflows by integrating SkyPilot and ClearML, reducing experiment setup time by 45%. Conducted A/B testing on fit algorithm improvements, resulting in 15% reduction in size-related returns.

Computer Vision Engineer at Motive (formerly KeepTruckin)

November 1, 2024 - August 27, 2025

Developed production-ready driver monitoring features detecting smoking and distraction behaviors with 94% accuracy. Architected multi-task YOLOx models with custom heads enforcing shared information between tasks. Reduced model size and inference time by 50% through 8-bit quantization using SNPE on Qualcomm devices. Delivered a semantic search system using OpenAI embeddings and Chroma vector store to support real-time, accurate answers across internal documents. Decreased data preprocessing pipeline latency by 60% through parallel processing. Created generic dataset package and automated labeling system using SAM, increasing annotation efficiency by 3x.

Software Engineer at Confiz Limited

February 1, 2022 - August 27, 2025

Junior Embedded Software Engineer at Vadion Pvt Limited

July 1, 2021 - August 27, 2025

Research Intern, VisPro Lab at Information Technology University

March 1, 2021 - August 27, 2025

Embedded Software Engineer Intern at Mentor - A Siemens Business

September 1, 2019 - August 27, 2025

Upgraded diagnostic tool for testing LIN and CAN-FD bus protocols in automotive applications. Redesigned transport layer for CAN-FD protocol supporting 8-Mbits/sec in Audi-Porsche mainboard.