Available · Q2 2026 Remote · Worldwide

Shaping AI that ships.

I'm Shahzaib Asif — an Applied AI/ML Engineer with 5+ years shipping production systems across computer vision, LLMs, MLOps, and edge AI. From cloud GPUs to silicon, I build the intelligence that actually makes it to customers.

Start a project See my work
05+ Years Experience
45+ AI Projects
$2.7M+ Cost Savings
PyTorch Computer Vision CUDA LLMs RAG Pipelines Docker LangChain Edge AI NVIDIA Jetson TensorFlow MLOps OpenCV PyTorch Computer Vision CUDA LLMs RAG Pipelines Docker LangChain Edge AI NVIDIA Jetson TensorFlow MLOps OpenCV
01 — At a Glance By the numbers

Five years. Real impact.
Measurable outcomes.

05+
Years Experience
45+
AI Projects
$2.7M+
Cost Savings
100K+
Images Analyzed
23+
Certifications
02 — About The engineer

Research shipped.
Not just published.

I work at the seam where experimental ML becomes operational software — taking models out of notebooks and into systems people depend on.

My work spans satellite imagery analysis, real-time video pipelines, and LLM-powered automation. I've shipped on cloud GPUs, on Raspberry Pis, and on everything between.

Based

Islamabad, Pakistan — working remote with teams worldwide.

Currently

Applied AI/ML Engineer at PreScouter Inc. (Remote, USA).

Focus Areas

Computer vision · LLM agents · MLOps · Edge deployment.

Distinctions

President's Gold Medal — M.Sc. CSE, NUST.

03 — Expertise Capabilities

Four disciplines,
one engineering practice.

Computer Vision

Object detection, segmentation, and classification across satellite imagery, real-time video, and embedded camera streams. From dataset design to deployment.

PyTorchOpenCVYOLORemote SensingSegmentation

Generative AI & LLMs

Production RAG pipelines, multi-step AI agents, and LLM-powered automation integrated with messaging APIs and backend services. Built for reliability, not demos.

LangChainRAGPrompt EngineeringAI AgentsTransformers

MLOps & Deployment

Containerized ML systems with reproducible deployments across multi-cloud and on-premise environments. Eliminating environment drift and the failures that come with it.

DockerCI/CDGitHubData PipelinesReproducibility

Edge AI & Optimization

Inference pipelines optimized for resource-constrained hardware. Real-time deep learning on NVIDIA Jetson, Raspberry Pi, and CUDA-accelerated GPU clusters.

NVIDIA JetsonRaspberry PiCUDAOptimizationParallel Computing
04 — Experience Trajectory

A timeline of systems
shipped to production.

10/2024 — Present
Now
Applied AI/ML Engineer
PreScouter Inc. — Remote, USA
→
  • Engineered LLM-powered automation pipelines integrated with messaging APIs and backend services, dramatically reducing manual research overhead.
  • Containerized ML systems using Docker for consistent, reproducible deployments across multi-cloud and on-premise environments.
  • Transformed experimental research prototypes into scalable, maintainable ML services with improved cross-team handoff quality.
02/2022 — 09/2024
Research Engineer (AI/ML)
NASTP, PAF Nur Khan
→
  • Designed end-to-end deep learning pipelines for object detection on large-scale satellite imagery, achieving improved geospatial detection accuracy.
  • Built scalable preprocessing, training, and deployment workflows for sensor-based datasets — enabling reliable operational ML pipelines.
  • Optimized inference pipelines for GPU and edge environments, reducing model latency on resource-constrained hardware.
  • Collaborated with software engineering teams to integrate ML models into production platforms.
01/2021 — 01/2022
Applied AI Engineer
Orel-Vision Pvt Ltd, NSTP Islamabad
→
  • Deployed optimized deep learning models on resource-constrained embedded devices, achieving real-time inference for edge-based computer vision.
  • Developed camera + GPS-tagged image acquisition systems for field-ready, location-aware dataset creation.
  • Streamlined vision model preprocessing pipelines, reducing inference latency in production deployments.
01/2020 — 01/2021
Research Assistant (AI/ML)
Image Analysis Lab, NUST Islamabad
→
  • Researched computer vision techniques for intelligent transportation systems, improving vehicle and traffic detection on real-world datasets.
  • Benchmarked deep learning detection architectures under operational constraints, informing model selection and deployment strategy.
Ongoing
Freelance AI/ML Engineer
Independent — Remote
→
  • Delivered end-to-end AI/ML applications with user-facing interfaces — proof-of-concept systems across diverse client requirements.
  • Deployed ML inference pipelines on Raspberry Pi and NVIDIA Jetson, enabling real-time edge AI on low-power devices.
  • Containerized ML solutions with Docker for reproducible cross-platform deployments.
05 — Selected Work Projects

Featured projects
worth noting.

Project / 001
0.94 0.89 0.96

Satellite Imagery Object Detection

Deep learning detection models for large-scale remote sensing analysis, with custom preprocessing and augmentation pipelines that improved geospatial detection performance.

PyTorchRemote SensingYOLO
Project / 002
SOURCE SWAPPED

Real-Time FaceSwap Pipeline

Containerized video processing pipeline enabling real-time AI face-swapping during live video conferencing, with stable meeting-time inference on standard hardware.

DockerOpenCVReal-Time
Project / 003
8x SPEEDUP CUDA + MPI

HPC Deep Learning Optimization

Optimized deep learning training using CUDA and MPI-based parallel computing — significantly accelerating model training across high-performance computing clusters.

CUDAMPIHPC
Project / 004
LLM PLAN TOOL MEM EXEC

Multi-Agent AI System

Production-grade agentic AI orchestrating planner, tool-use, memory, and executor sub-agents. Autonomous task decomposition across enterprise workflows with human-in-the-loop guardrails.

LangGraphOpenAIAgents
Project / 005
QUERY KNN k=5

Enterprise RAG Platform

End-to-end retrieval-augmented generation system with hybrid search, vector embeddings, re-ranking, and citation tracking. Serves millions of queries across private knowledge bases.

PineconeLangChainEmbeddings
Project / 006
FUSION OUT

Multimodal Understanding

Vision-language model fine-tuned for cross-modal reasoning over images, text, and audio. Powers document AI, visual QA, and accessibility features for enterprise clients.

CLIPVLMFine-Tuning
06 — Credentials Education

Trained academically,
tested in production.

M.Sc. Computational Science & Engineering
National University of Sciences and Technology (NUST)
Islamabad, PK · GPA 3.9/4.0 President's Gold Medal
B.Sc. Electrical Engineering
COMSATS University Islamabad — Lahore Campus
Lahore, PK · GPA 3.3/4.0 Runner-up Overall
Selected Certifications
  • Generative AI Language Modeling with Transformers IBM · Coursera
  • AI Agents Using RAG and LangChain IBM · Coursera
  • Building Systems with the ChatGPT API DeepLearning.AI
  • PyTorch Essential Training: Deep Learning LinkedIn Learning
  • Edge Computing for Computer Vision MachVIS · SEECS NUST
  • TensorFlow World Extended Islamabad GDG Cloud Islamabad
07 — Contact Start a project

Let's build
something unforgettable.

Got a project, a research problem, or a system that needs to scale? Send an inquiry below and I'll respond within 24–48 hours.

LinkedIn
/in/shahzaib-asif
Fiverr · Vetted Pro
AI integration services
Project Inquiry Typical response within 24 hours
Under $500 $500 – $1,500 $1,500 – $5,000 $5,000 – $15,000 $15,000+ Open to Discussion
Sent inquiries are treated with strict confidentiality.
Message sent! I'll be in touch within 24—48 hours.
Something went wrong. Please try again or reach out via LinkedIn.
Limited Offer
Let's start now & get 27% off Send Inquiry