Muhammad Shahmir Zaman
AI EngineerAI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
AI Engineer with 1+ years in Computer Vision, NLP, and LLMs
AI / Machine Learning Engineer with production experience in Computer Vision, Natural Language Processing, and Large Language Model (LLM) deployment. Proficient in PyTorch, Hugging Face Transformers, LangChain, and LangGraph for building scalable, full-stack ML pipelines and agentic AI systems. Hands-on experience with RAG architectures, multi-agent frameworks, model fine-tuning, quantization, and MLOps workflows including gRPC-based microservices. Proven ability to deliver measurable business impact — improving model accuracy, reducing inference latency, and accelerating deployment cycles in fast-paced, production environments.
FAST-NUCES
Bachelor of Science · Computer Science
August 1, 2021 – June 30, 2025
QBS Co
AI Software Engineer
September 1, 2025 – Present
Karachi, Sindh, Pakistan
Intelik
AI Engineer
June 1, 2025 – September 1, 2025
Karachi, Sindh, Pakistan
Abandoned Object Detection System
January 1, 2025 – December 31, 2025
Developed a production-ready Abandoned Object Detection system using Python, OpenCV, and YOLO combining real-time object detection, segmentation, and pose estimation with background subtraction to identify unattended items in complex indoor environments. Devised a hybrid centroid-based tracking algorithm (IoU + distance metrics) and multi-state ownership classifier (confirming → owned → warning → abandoned) to accurately link objects to persons and track state transitions across frames. Designed a Locked Zone suppression mechanism with automatic pickup detection and furniture-filtering baseline to eliminate false positives from static scene elements and transient human-object occlusions. Exposed detection pipeline over gRPC, enabling low-latency streaming of alert events to connected monitoring dashboards and security systems.
Text Summarization with Fine-Tuned Transformers
January 1, 2025 – December 31, 2025
Fine-tuned T5-base on the CNN/DailyMail dataset for abstractive text summarization; evaluated using ROUGE-L metrics to benchmark generation quality. Applied beam search decoding and length-penalty tuning to improve summary coherence for downstream NLP applications.
Multi-LLM Ensemble System with Reinforcement Learning
January 1, 2025 – December 31, 2025
Built a deployment-grade multi-agent LLM system integrating OPT-125M and Falcon-7B with a PPO-based reward model and Knowledge Transfer Protocol for improved response coherence. Applied 4-bit quantization and optimized the distributed inference stack, reducing latency while maintaining output quality on both GPU and CPU. Structured for flexible deployment via LangGraph agentic orchestration framework.
Early Alzheimer's Diagnosis via Deep Learning (FYP)
January 1, 2024 – December 31, 2025
Trained CNN and ensemble deep learning models on 5,000+ MRI scans for binary and multi-class Alzheimer's disease classification, achieving 0.723 AUC-ROC. Leveraged Kernel PCA (KPCA) for dimensionality reduction — achieving 65% feature compression while preserving 98% of variance improving downstream model efficiency. Led a 3-member research team; evaluated multiple ML baselines (Random Forest, SVM, XGBoost) before selecting deep learning architecture for best generalization.
House Price Prediction - End-to-End ML Pipeline
January 1, 2024 – December 31, 2024
Developed a regression model using Random Forest and Gradient Boosting (XGBoost), achieving R2 = 0.89 on held-out test data after feature engineering and cross-validation. Deployed model as a REST API using Django; containerized with Docker for reproducible deployment.
Foundations of Data Science
Coursera / Google
January 1, 2024 – Present
Cultural Fit Analysis
The candidate's diverse project portfolio, spanning computer vision, NLP, and LLMs, indicates a broad interest and adaptability to various AI challenges. Experience in both professional and academic settings, including leading a research team, suggests a collaborative and growth-oriented mindset. The explicit mention of improving business metrics (e.g., reducing hallucination rate, cutting deployment time) aligns with a results-driven culture. The candidate's skills and project types are highly relevant to an AI Engineer role, demonstrating a strong cultural fit for a technically demanding and innovative environment.
Soft Skills & Operational Fit
The candidate's project descriptions and professional experience highlight strong problem-solving skills, evidenced by devising hybrid tracking algorithms and locked zone suppression mechanisms. Leadership is shown in leading a research team. The ability to collaborate with C-suite stakeholders and achieve 100% on-time delivery indicates strong client management and operational reliability. The focus on MLOps and CI/CD suggests an understanding of efficient, reproducible deployment workflows.