Basit A.Private profile: Upwork shows its earnings only to signed-in members, so we check it every few weeks, not daily. MRR here is a rough estimate (Upwork's six-month earnings ÷ months active). Last read Nov 5, 2026.

I am a Machine learning engineer specializing in NLP with over four years of experience working with multinational organizations to small-scale startups, working on Machine learning, Deep learning, NLP, computer vision, predictive modeling, financial modeling, and many more. 1- At IBM, I was responsible for conducting data analysis, visualization, cleaning, modeling, and testing as part of the data analytics team. I handled textual data by deploying sophisticated deep learning and natural language processing algorithms. Bettered existing language modeling systems and ran multiple experiments through state-of-the-art predictive models such as IBM Watson, LSTMs, fastText, Elmo, BERT, Transformer networks, and fast.ai. 2- At Trillo, I was building end-to-end pipelines from the R&D of a new ML/DL-based project to data cleaning, visualization, modeling, testing, and deployment using flask/Pytorch serving in a docker container deployed on a google cloud platform. Trillo is a small startup of 6-7 people working as a google cloud partner( and several other clients providing machine learning, AI, and DevOps solutions). 3- I have built the most popular machine learning and data analysis classifiers from the ground up myself, no libraries or anything. This gives me a tremendous insight into which is the suitable algorithm for your job. 4- I have worked on a supply-demand project, which was being run by Kaggle, which is one of the biggest data analytics firms; for its client UBER. My work was ranked 55th (Top 10 %). The goal was to control surge pricing, i.e., the pricing factor, by predicting which area would have higher demands. Work history : - Fintech-based classification, summarization, and GLUE tasks by training Roberta and FinBERT architecture-based language models from scratch and fine-tuning existing models with over 30Million corpus size. - Recently built a fully automated stock market analysis platform from scratch. It had many modules which allowed for automated trading. A cascade made decisions by the platform of Machine learning classifiers. It had the ability to analyze good or bad trades using SMA's and Stochastic indicators, as well as many other conditions. Analysis was done using visual inputs as well as numeric data. - Google cloud assistant mobile and web-based application using Dialogflow and deep learning models for sales forecasting for a marketing organization having over 1500 stores to deliver goods all across the United States. - Advanced ticket routing automation by building a text classification system from scratch; utilized fast.ai’s universal language as transfer learning and added a classification model as its head. It achieved a record score of 0.89 F1 on a test set using ULMFiT. - Open-domain question answering system based on SOTA Bert-QA, Bert reranking, and bm25 on 1M+ articles.