Muqsit Amir Email & Phone Number

Lahore

I specialize in processing agent-caller interaction data to create predictive models that optimize agent-caller pairings in real-time, maximizing the probability of successful sales. My role involves the day-to-day analysis of real-time interaction data to identify and address anomalies in agent or caller behavior, ensuring that my models adapt to these irregularities. I closely monitor the incremental revenue generated by my models through intelligent routing and run diagnostics on underperforming models to fine-tune them for better results.My work relies heavily on advanced techniques such as Bayesian Statistics, 1 and 2 parameter logistic models, and MCMC estimation methods. I'm responsible for maintaining AI models in production for a European client, using Afiniti's ExperienceAIElite product. My focus is on analyzing performance and retraining the company's core AI pairing models to ensure positive revenue in dynamic environments.In addition, I build efficient data pipelines to ensure the right data is used for pairing, and devise modeling strategies to capture ever-changing trends in real-time, ensuring continuous positive revenue inflow. My expertise includes using platforms and tools such as MySQL, Python, R, and Keras to support and enhance these efforts.Modeling human behavior with AI to ensure that customers are paired with the right agent to maximize revenue is at the core of what I do.

Lahore, Punjab, Pakistan

At LUMS, I fulfill a multifaceted role as a Full-Stack Developer, working across the entire technology stack from frontend development with React to backend development with Django and Django Rest Framework. My responsibilities include overseeing the full deployment process, configuring Nginx, managing Linode servers, and ensuring efficient database management.I am deeply involved in innovative computer vision projects, including big cat classification and remote forest fire detection. My work extends beyond coding; I oversee and manage the software on remote cameras, optimizing their performance and reliability. A notable achievement is fine-tuning the MegaDetector model on a custom dataset, which has improved accuracy in classifying animals like leopards while reducing false positives from environmental factors.Additionally, I have deployed machine learning models like TinyYOLO on Jetson machines, enabling real-time object detection and enhancing the capabilities of remote cameras. I’m also exploring the integration of multimodal Large Language Models (LLMs) like CLIP and OWL for wildlife and fire detection using text prompts, pushing the boundaries of real-time environmental monitoring.Our team's work was recently recognized at the Asia-Pacific Triple E Awards for advancing education and environmental conservation through technology.In summary, my role at LUMS blends frontend and backend development, overseeing deployment orchestration, database management, and cutting-edge computer vision and machine learning applications, allowing me to contribute significantly to research projects and continuously expand my technical skills.

Lahore, Punjab, Pakistan

As a freelance Data Extraction engineer working under SoftCreed on Upwork, I used Python's Scrapy framework and Selenium for web automation helping them extract and clean data from complicated dynamic websites for their usage. I worked for a year on industry-grade data pipelines and scraped dynamic websites for global clients, including American and Irish companies.