Mansoor Shah
About
Detail
Machine Learning Engineer
Pakistan
Contact Mansoor regarding:
work
Full-time jobs
Starting at
USD3K/month
groups
Networking
Résumé
Jobs
verified_user
0% verified
Machine Learning EngineeredgeJan 2026 - Current (4 months)- Machine Learning EngineerVisNext Software SolutionsApr 2025 - Jan 2026 (10 months)- Developed and implemented a comprehensive feature reduction pipeline using techniques such as NA filtering, variance thresholding, Median Absolute Deviation (MAD), and clustering to optimize peptide data for machine learning models. - Engineered a robust data preprocessing workflow for peptide datasets, incorporating advanced techniques like power transformation and robust scaling to handle skewed data distributions, enhancing model performance and accuracy. - Designed and executed a machine learning model training framework utilizing AutoGluon for binary classification tasks, achieving high accuracy and ROC-AUC scores, and conducted detailed model evaluation using custom metrics and SHAP analysis for feature importance insights.
- Machine Learning EngineerDiveDeepAIApr 2023 - Mar 2025 (2 years)- Applied prompt engineering techniques to fine-tune LLM behavior in RAG pipelines, improving document retrieval accuracy and conversational coherence. - Coordinated with clients to understand AI needs and delivered tailored NLP solutions using LLMs, RAG systems, and Prompt Engineering, integrating ElasticSearch and LangChain for document retrieval and interaction flow. - Built production-ready AI models with Python, PyTorch, and Hugging Face. - Automated data entry processes via Selenium scrapers, achieving a 50% speed increase and fewer manual errors, seamlessly integrated into a FastAPI backend. - Developed a content-generation agent using ChatGPT, Gemini, and Claude, reducing creation time by 40%. - Fine-tuned LLMs using LoRA and QLoRA
Education
verified_user
0% verified
- Bachelor's degree, Computer Software EngineeringCOMSATS University IslamabadJan 2018 - Dec 2022 (5 years)Main Courses Studied: • Data Science • Design Patterns • Data Structure & Algorithms • Operating Systems (Ubuntu) • Communication & Networking • Statistics and Probability • Artificial Intelligence • Object Oriented Programming • Database Systems
Projects (professional or personal)
verified_user
0% verified
- BBRAIN TUMOUR DETECTION FROM MRI USING CNN● Designed & Developed a CNN-based deep learning model achieving 91% accuracy in brain tumor detection. ● Trained the model on a dataset of 3,000 MRI scans after applying some data augmentation techniques to artificially enlarge the dataset. ● Applied data augmentation techniques for improved generalization on unseen data. ● Optimized CNN architecture to reduce processing time by 30%. ● Deployed the model as a web application using Streamlit, facilitating seamless tumor detection for medical professionals.
- SSPEECH EMOTION RECOGNITION● Developed a state-of-the-art speech emotion recognition model using CNN architecture with 85% accuracy. ● Trained the model on a dataset of over 20,000 audio samples, achieving a 25% improvement in emotion recognition accuracy compared to traditional methods. ● Extracted emotion-related features such as pitch, speech speed, & tone using spectrogram analysis & advanced signal processing techniques. ● Optimized the model for real-time performance, enabling it to process speech samples within milliseconds.