We can organize an interview with Aldin or any of our 25,000 available candidates within 48 hours. How would you like to proceed?
Schedule Interview NowMy name is Mansoor T. and I have over 1 years of experience in the tech industry. I specialize in the following technologies: Generative AI, Prompt Engineering, LLM Prompt Engineering, ChatGPT, Natural Language Processing, etc.. I hold a degree in Bachelor's degree. Some of the notable projects I’ve worked on include: AI Assistant, Urdu Image Captioning, Text to SQL LLM, Vegetable Price Forecast, AI-driven 2D-to-3D Modeling in Blender, etc.. I am based in Lahore, Pakistan. I've successfully completed 10 projects while developing at Softaims.
I thrive on project diversity, possessing the adaptability to seamlessly transition between different technical stacks, industries, and team structures. This wide-ranging experience allows me to bring unique perspectives and proven solutions from one domain to another, significantly enhancing the problem-solving process.
I quickly become proficient in new technologies as required, focusing on delivering immediate, high-quality value. At Softaims, I leverage this adaptability to ensure project continuity and success, regardless of the evolving technical landscape.
My work philosophy centers on being a resilient and resourceful team member. I prioritize finding pragmatic, scalable solutions that not only meet the current needs but also provide a flexible foundation for future development and changes.
Main technologies
1 year
1 Year
1 Year
1 Year
Potentially possible
Careem
This project allows users to upload and discuss documents using AI models like OpenAI, Claude, and Gemini. Documents are securely stored in an S3 bucket, with metadata managed in MongoDB. The system selects the best AI model based on document length for accurate responses. Users can ask questions, and the AI provides context-aware answers. This ensures efficient, secure, and cost-effective document management and interaction, making it ideal for handling legal, business, or general inquiries with precision and ease.
Our Task was to make a model that takes in Image and describe the image in Urdu Language. We created a novel dataset of Pakistani 1000 images with 5 Urdu captions per image. Models we used were 1) CNN-LSTM (Res-net, VGG16) 2) Blip Transformers We achieved a great Accuracy (Bleu-1 score) of 0.75 which was on par with latest research papers. Finally we deployed the model using flask and made an android app for it. We plan to publish our findings and dataset in a research paper.
Developed a data processing tool using LangChain’s SQL agent, designed to simplify complex database interactions. First, data is retrieved from the company’s API, then loaded, cleaned, and organized by a custom DataExtractor class, which stores it in both SQLite and CSV formats. Built a QueryExecutor to run SQL queries and return results in a structured JSON format. Integrated LLaMA 3.1 70B model for natural language SQL generation, so users can input simple instructions that convert into complex SQL queries—making it easy to analyze large datasets and generate custom reports.
I worked on a project predicting prices of 20 vegetables in the Pakistani market. After conducting EDA and addressing missing values, I incorporated external factors like petrol prices, USD exchange rates, and weather conditions to enhance prediction accuracy. I explored time series methods, including ARIMA and Facebook Prophet, and implemented advanced deep learning models like LSTMs and transformers, specifically the i-transformer, achieving MAE: 0.090, MSE: 0.026, and RMSE: 0.162. This project aims insights for farmers and government agencies on managing supply and demand.
This project uses the Stable Fast 3D AI model to transform 2D images into 3D character parts, which are then combined in Blender using Python scripting. The script automatically imports the character's body and face models, aligns them, rotates the face, and scales it for a natural fit. By setting precise positions and adjustments, this project quickly builds a complete 3D character ready for animation. This setup makes it easy to go from a 2D design to a fully assembled, animated 3D model in Blender.
Bachelor's degree in Data Science