● Conducted in-depth RnD and gathered system requirements in order to solve the problem of vehicle categorization for the biggest network of highways of Pakistan i.e the Motorways of Pakistan.● Gathered, cleaned and annotated dataset of more than 100k image to be used in training of AI.● Created pipelines and wrote rules and techniques for data wrangling i.e data cleaning, data processing, data annotations and data augmentations.● Played a leading role in training and implementing Machine Learning Models and Computer Vision techniques on Dahua cams in order to solve the vehicle categorization problem. Some of the tools and technologies used to achieve that were PyTorch, Tensorflow, Numy, OpenCV, Supervision, Roboflow, Pandas and NVIDIA CUDA/CuDNN etc● The problem was solved by training Artificial Intelligence Model with an accuracy of 95% and above.● Lead and trained a team of interns and oversaw designated workflows and tasks and achieved maximum throughput and achieved the goals set by the organization.● Worked on the optimization of the ML Model by performing Int8 Quantization from Int32 by utilizing tools such as ONNX and OpenVino. As a result the model utilized 80% less physical resources and ended the reliance of inference on Nvidia GPUs. Instead the inference was made depended on common commercial use Intel CPUs thus reducing the cost of deployment significantly.● Reported every single milestone to managers and presented the first version of the final product to all the major stakeholders, not only meeting their satisfaction but also surpassing their expectations.● Currently in the phase of deployment of the AI to all the major areas of Motorways and conducting thorough testing in order to make sure the entire system is bullet proof.

● Collaborated with a team of engineers in optimizing backend infrastructure for their next generation LMS portals.● Collected client side requirements by effectively engaging with the customers and created effective work flows surrounding said requirements.● Collected, mined and analyzed user data and helped build multidimensional snowflake database warehouse schema using SQL and PL/SQL.● Performed analytical processing on user and customer data using Hybrid Online Analytical Processing (HOLAP) and effectively stored them in fact tables.● Helped backend and frontend teams with the integration of the data warehouse using Python and JavaScript.● The overhauled backend resulted in a highly scalable portal for educational institutions of all sizes and increased our target audience by 10% and increased our revenue by 30%.● Effectively participated in team building activities including helping newcomers onboard and getting them familiarized with current technologies being used and the processes being carried out.

● Collaborated with a team of senior electrical engineers by utilizing Python and implemented ensemble learning methods for classification and regression discussed in multiple research papers and utilizing them on data gathered using EEG machines and created a program that helps Cerebral Anoxia(Paralysis) patients communicate using basic auditory commands● Wrangled 10GBs of structured data in CSV format of electromagnetic brain waves using EEG machines and trained multiple machine learning models on Raspberry Pi, effectively enabling it to predict with accuracy of around 90%.● Presented the findings to the team lead and drafted a report detailing the effectiveness and value proposition if this product is further refined and rolls into a testing phase.● Prepared an extensive report detailing all the research and technologies used which provided the said results for all the stake holders involved in this project.