Islamabad, Islāmābād, Pakistan
A Senior AI Engineer/Consultant focused on building LLM-powered systems, AI agents, ML pipelines, and production-grade AI products. Over the past 7+ years, I’ve worked across NLP, computer vision, document intelligence, anomaly detection, recommender systems, RAGs, agentic flows, and cloud-based ML deployments. Currently at QuickStart Inc., I work on AI-driven education products including, personalized learning systems, career path recommendations, interview/assessment tools, and LLM-powered course experiences. Previously at Red Buffer, I led ML teams, managed client-facing AI products, mentored engineering teams, and contributed to production systems across healthcare, fintech, sports analytics, OCR, and cloud cost intelligence. I’m AWS Machine Learning Specialty certified, AWS x3/GCP x1 certified, and especially interested in LLMOps, agentic workflows, scalable AI systems, and turning AI prototypes into reliable products.
- Leading the architecture of scalable LLM-powered learning intelligence systems for personalized course generation, skill-gap analysis, and career-path recommendations across IT and technology domains. - Designing agentic AI workflows, RAG pipelines, and learner-context orchestration systems for adaptive learning, automated assessments, and AI-powered interview simulations. - Evaluating, integrating, and productionizing third-party AI APIs, foundation models, and LLM services across multi-product EdTech ecosystems. - Advising product, engineering, and leadership teams on AI strategy, solution architecture, LLMOps, feasibility analysis, and production-grade implementation roadmaps.
- Conducting a research project focused on optimizing the accuracy of a model for detecting digitally manipulated images and videos. The approach involves employing an ensemble of various CNN models to enhance the effectiveness of the detection process. - Involved in partially providing services for the deployment, maintenance, and marketing of a product that creates artfully inspired AI-generated photos and videos. - Managing a team to build a product that solely focuses on in-depth explaining and generating flowchart visualization of a code base. Providing line-by-line explanations and making suggestions on how to further optimize the snippet. - As the product lead, managed a real-time solution using generic cameras to detect ball speed and count the occurrences of a brand logo in domestic cricket matches. Continual research efforts were dedicated to upgrading the vision and tracking models for improved near-to-real-time performance. - Managing an anomaly detection project that uses statistical methods to detect pattern-based anomalies with a hierarchy of tags. The model was trained for financial data of AWS and Azure cost files. It also caters to the edge cases like retagging of names. The end goal is to allow the sales team a new stable environment where they could showcase new/custom features. The code is on PySpark and the databases used are ClickHouse and PostgreSQL. Project orchestration is on Kubeflow. - Actively participating in the hiring process and setting up an exception ML team. - Managing ML internship program.
- Supervising and catering the ML part of a project for an Australian vend to extract information from all identification documents of the user and electricity bills to provide a better saving plan. (Image segmentation, Image pre-processing, Text post-processing, Regex, Key word base rules, OCR) - Deployed a module for a client on AWS EC2, a US-based project to develop an engine the extracts core content (tables, images, and text) from the information extracted through the document. All major media types like PDFs, text files, and web pages were handled. (Text Classification, Computer Vision, Web scrapping, Abiword, Tabula, HTML) - Met requirement for a US-based company to developed and deploy an AWS Lambda function that interacts and dispenses products from a custom-built vending machine. (SOAP, AWS, FlexConnet, Raspberry Pi 2) -Trained a model to extract vital information from driving licenses of different states of the US for a client with an accuracy of 85% and above. (NLP) - Completed a project to develop a bot for crypto-trading and applying multiple techniques to test different strategies like, Buy and Hold, MA, etc on past data. (Forecasting)
- Leading a small team of interns to an efficient and systematic cleansing of project data.
- Developing a URL Shortening web application as a part of internee developer test. - Updating existing Welcome and Subscription emails for Quixel Subcrised customers. - Developing a client that hooked up with Quixel API and allowed flexible download of library items for internal use. - Creating email templates for Quixel newsletter
Taaleem Foundation strives to educate a social change through quality education in the under privileged areas of Pakistan, especially Balochistan. During my tenure at the organisation I got appreciated in the following domains: - Website analyst for traffic growth. - Strategies for company advertisement. - Designing company brochures and calendars. - 3D designing - Designing Company Logos - Visual Effects