Zeeshan Javed
@zeeshanjaved1
Senior AI and computer vision engineer, 5+ years shipping production vision, deep learning, and LLM systems for international clients.
PakistanWhat I'm looking for
I design, build, and deploy production AI systems across computer vision, deep learning, and LLM applications, with 5+ years of experience delivering for international clients.
As founder of VisAItech, I work independently with clients across the UK, US, and Europe and hold a 100% Job Success Score on Upwork, scoping engagements end to end from requirements through architecture, implementation, deployment, and handover documentation. Delivered work includes an automated golf shot detection pipeline (pose-driven state machine, custom-trained YOLOv8 ball tracker, audio-based impact validation), a FaceNet dwell-time analytics system with automated reporting, video forensics analysis, and OCR fine-tuning.
At KICS-CLE, UET Lahore, I am Assistant Manager Research. I built an automated video-analysis platform that runs batch inference over a year of Urdu broadcast news into a queryable SQL database, and led generative AI work on diffusion-based sign-language video synthesis with quantitative quality evaluation (face-region SSIM, optical-flow warping error). I also build GPU-accelerated multi-GPU training pipelines using CUDA and cuDNN.
I was a Visiting Researcher at the University of Konstanz, Germany, working on OCR, transcription, and annotation methods for historical low-resource scripts. My paper on a diachronic Urdu text and image corpus is accepted at LREC 2026, and a Diffusion-GAN framework for Pakistan Sign Language production is under review at IEEE Access.
Core stack: Python, PyTorch, TensorFlow, OpenCV, Detectron2, MediaPipe, YOLO, Stable Video Diffusion, LangChain, FAISS, Whisper, SQL, Git, Linux.
Fluent in English and experienced in fully remote, distributed collaboration across time zones. Portfolio: mzeeshanjaved.com
Experience
Work history, roles, and key accomplishments
Lead R&D in computer vision, document AI, and generative AI. Built and deployed an automated video-analysis platform that runs batch inference over Urdu broadcast news into a queryable SQL database. Lead diffusion-based sign-language video synthesis with quantitative quality evaluation (face-region SSIM, optical-flow warping error). Build multi-GPU training pipelines with CUDA and cuDNN.
Deliver production computer vision and AI systems to international clients across the UK, US, and Europe, with a 100% Job Success Score on Upwork. Shipped an automated golf shot detection pipeline (pose-driven state machine, custom-trained YOLOv8 ball tracker, audio impact validation), a FaceNet dwell-time analytics system, video forensics, and OCR fine-tuning. Scope engagements end to end.
Designed a course on digitisation workflows for low-resource languages and investigated OCR, transcription, and annotation methods for historical Urdu scripts. Collaborated with international researchers on language resources and digital corpora at the intersection of language technology, digital preservation, and applied AI.
Developed deep learning pipelines for text localization, recognition, and historical document processing in low-resource Urdu, using GAN-based augmentation to improve robustness. Built the Urdu News Video Analytical System: a self-built YOLO ticker-localization dataset, GAN-augmented OCR, and in-house ASR, speaker identification, and summarization integrated into production web systems.
Research and development in image processing and deep learning: text detection and recognition experiments for Urdu document analysis, dataset preparation, and model training and evaluation supporting the centre's document AI work.
Education
Degrees, certifications, and relevant coursework
University of the Punjab
Master of Philosophy, Artificial Intelligence
2022 -
Grade: 3.3/4.0
Coursework completed (CGPA 3.3/4.0); thesis defense pending. Thesis: Smart Multi-Source Video Content Retrieval, a RAG system for semantic search over video using Whisper ASR, EasyOCR, LangChain and FAISS with an LLM query interface. Coursework: Deep Learning, Mathematics for AI, NLP, Reinforcement Learning, Data Science for Assistive Health.
University of Narowal
Bachelor of Science, Information Technology
2016 - 2020
Grade: 3.23/4.0
Final year project: a person authentication system using deep neural networks, funded by the Government of Pakistan (IGNITE).
Availability
Location
PakistanAuthorized to work in
Website
mzeeshanjaved.comPortfolio
mzeeshanjaved.comSalary expectations
Social media
Job categories
Interested in hiring Zeeshan?
You can contact Zeeshan and 90k+ other talented remote workers on Himalayas.
Message ZeeshanGet matched with your dream remote job
Sign up now and join over 250,000+ remote workers who receive personalized job alerts, curated job matches, and more for free!
