Skip to content
Botmer International®
Botmer talent profile

Moazam Adrees

Senior Deep Learning Engineer, Computer Vision

Deep learning engineer focused on computer vision, multimodal language models, remote sensing, and anomaly detection. Builds systems for transmission inspection, streaming security, industrial safety, and retail analytics. His work combines high-resolution imagery, depth estimation, 3D reconstruction, sparse-label training, RAG guidance, and efficient model adaptation.

4+ years
production experience
Computer vision & MLLMs
primary engineering focus
Sparse-label learning
demonstrated model specialization

Lahore, Pakistan

Meet the engineer

AI engineering with production depth

Where Moazam adds value

  • Develops inspection systems for high-resolution drone imagery, addressing tiling and overlap alignment and estimating cable sag, pole height, and lengths.
  • Uses MoGe and Metric Depth models for 3D reconstruction from 2D images, alongside multi-GPU training and on-the-fly image augmentation.
  • Builds multimodal RAG guidance and fine-tunes Qwen vision-language models for defect grounding and real-time detection using sparse labeled datasets.
  • Delivers anomaly detection, few-shot satellite analytics, PPE monitoring, and retail tracking using temporal models, Vision Transformers, transfer learning, and YOLO.
Explore professional experience
Core competencies

AI systems grounded in software engineering

01

Computer vision and geometry

Builds detection, segmentation, depth-estimation, and 3D reconstruction workflows for high-resolution imagery, using tiling, alignment, monocular vision, object tracking, and multi-GPU training for industrial inspection applications.

  • OpenCV
  • MoGe
  • Metric Depth
  • YOLOv8
  • PyTorch
  • 3D reconstruction
02

Multimodal models and retrieval

Combines multimodal language models and retrieval-augmented guidance with domain manuals, zero-shot labeling, grounding tasks, and sparse-data fine-tuning to support real-time asset inspection and defect detection.

  • Qwen 3.5-VL
  • RAG
  • MLLMs
  • Grounding
  • Zero-shot learning
03

Anomaly and sparse-label learning

Uses temporal CNN and Transformer models, few-shot learning, Vision Transformers, and transfer learning to handle drift, class imbalance, limited annotations, remote sensing, and abnormal behavior detection.

  • Transformers
  • CNNs
  • Few-shot learning
  • Vision Transformers
  • EfficientAD
04

ML implementation and efficiency

Works with Python, PyTorch, TensorFlow, Keras, Linux, Docker, and cloud platforms, with research comparing small and large language models for lower-carbon industrial AI workloads and deployment.

  • Python
  • PyTorch
  • TensorFlow
  • Docker
  • AWS
  • Green AI
Experience

From models to product delivery

Jul 2024 – PresentMeriti Inc

Deep Learning Specialist

Develops transmission-inspection, multimodal guidance, streaming anomaly detection, and remote-sensing systems. Handles high-resolution imagery and sparse labels using geometry models, multi-GPU training, vision-language fine-tuning, temporal networks, retrieval, and few-shot learning methods across production workflows.

  • Built 3D measurement pipelines from 100-megapixel drone imagery.
  • Fine-tuned Qwen 3.5-VL on approximately 1,000 defect images.
  • Reduced abnormal-login false positives by over 25%.
  • Delivered agricultural remote-sensing analytics with fewer than twenty labels per region.
Mar 2023 – Jun 2024HCRL, UET

Machine Vision Engineer

Designed and deployed vision systems for industrial safety and retail analytics. Applied transfer learning to PPE detection and developed object tracking and heatmaps for store analysis, evaluating models against defined image datasets and quality metrics.

  • Built Safe Sense PPE compliance monitoring with mAP 0.93.
  • Created Retailify using YOLOv8 tracking and heatmaps.
  • Achieved F1 0.97 for Retailify on a 5,000-image dataset.
Selected work

AI systems built for practical use

Highlights from professional projects, product delivery, and specialist work.

Meriti Inc

Transmission infrastructure inspection

Developed high-resolution drone-image analysis with grid tiling, overlap correction, MoGe and Metric Depth reconstruction, and multi-GPU augmentation. Built sparse-label defect grounding and multimodal RAG guidance using job aids, technical texts, and inspector manuals.

  • PyTorch
  • MoGe
  • Qwen 3.5-VL
  • RAG
Meriti Inc

Abnormal streaming login detection

Built an anomaly-detection system for a streaming platform handling millions of daily logins. Combined temporal CNN and Transformer models to address data drift and class imbalance, reducing false positives by more than 25%.

  • CNNs
  • Transformers
  • Anomaly detection
Named project

Safe Sense

Developed PPE compliance monitoring using transfer learning on the CHV dataset. The system detected safety violations with 87% detection and mAP 0.93, supporting industrial safety inspection and demonstrating potential for industrial deployment.

  • Transfer learning
  • Computer vision
  • PPE detection
Named project

Retailify

Built a retail analytics platform combining YOLOv8 object tracking with heatmaps for store-layout and sales decisions. Evaluated the system on a 5,000-image dataset, achieving F1 0.97 for its tracking and analytical workflows.

  • YOLOv8
  • Object tracking
  • Heatmaps
Education & professional focus

A foundation for practical delivery

Education & qualifications

M.S. in Artificial Intelligence (in progress)

University of Engineering & Technology, Lahore

2026

BSc. in Mechatronics & Control Engineering

University of Engineering & Technology, Lahore

2024

Professional focusVision systems under limited labels

His work emphasizes sparse-data fine-tuning, transfer learning, zero-shot and few-shot methods, and multimodal grounding for inspection and remote-sensing systems with constrained annotations.

Professional focusPractical and efficient industrial AI

His projects connect geometric measurement, defect detection, safety monitoring, retail analytics, and anomaly detection with research into efficient small-model alternatives for industrial tasks.

Talk to Botmer

See if Moazam fits your AI engineering team

Share your requirements, current challenges, and delivery goals. Botmer will review the brief and help arrange the next step.

Request a talent match