Amar Sohail

Services

What I can build for you.

From first prototype to production scale — I design, build and run the whole system, not just the model.

Generative AI & LLM Apps
Chatbots, copilots, search and assistants grounded in your own data — production RAG with the right chunking, retrieval and evaluation, so answers stay accurate as your data grows.
- LangChain
- LlamaIndex
- OpenAI
- Claude
- Gemini
- Pinecone
Agentic AI & Automation
AI agents and multi-agent systems that plan, use tools and finish real work — with memory, guardrails and human-in-the-loop review where it matters.
- LangGraph
- CrewAI
- AutoGen
- MCP
- Claude Code
- n8n
AI Architecture & Strategy
The right models, data flows and infrastructure before you build: architecture reviews, build-vs-buy and model selection, evaluation, cost and latency planning, and a roadmap your team can execute.
- Model selection
- Evals
- Guardrails
- Cost & latency
- Roadmaps
Model Training & Fine-tuning
Models trained on your data: dataset curation, labelling and synthetic data, LoRA/QLoRA fine-tuning of LLMs and diffusion models, GPU training runs and rigorous evaluation.
- PyTorch
- Hugging Face
- LoRA/QLoRA
- TensorFlow
- Datasets
- Synthetic data
Image & Video Generation
Automated pipelines for AI images, video and voice: Flux and SDXL generation, LoRA training for consistent characters, video generation and voice cloning — at production scale.
- ComfyUI
- Flux
- SDXL
- LoRA training
- Video generation
- Voice cloning
Computer Vision
Image and video understanding: object detection, segmentation, classification, OCR and visual inspection models, optimized for fast, reliable inference.
- PyTorch
- OpenCV
- Detection
- Segmentation
- OCR
NLP & Speech
Text classification, entity extraction, summarization, semantic search and speech-to-text — from classic NLP to LLM-powered pipelines for documents, support and knowledge bases.
- Transformers
- Embeddings
- NER
- Whisper
- Semantic search
Data Science & Datasets
Turn raw data into models and decisions: data pipelines, dataset creation and cleaning, feature engineering, forecasting, recommendations and analytics.
- Python
- pandas
- scikit-learn
- SQL
- ETL pipelines
MLOps & GPU Infrastructure
Deploy, scale and monitor models in production: GPU inference, model serving APIs, CI/CD for models, drift monitoring and cost control on Kubernetes across AWS, Azure and GCP.
- Kubernetes
- Docker
- MLflow
- SageMaker
- FastAPI
- VRAM optimization

Amar Sohail is a freelance AI engineer and AI architect. He builds generative AI and RAG applications, agentic AI and multi-agent systems, trains and fine-tunes models on custom datasets, delivers computer vision, NLP and data science solutions, builds image and video generation pipelines, and runs models in production with MLOps and GPU infrastructure. He is available for freelance AI engineering, consulting and contract work worldwide.

Discuss your project

Free 30-minute call. No commitment.

Portfolio

Selected work.

AI products and the platforms behind them — from generative AI content pipelines to knowledge retrieval at scale.

Testimonials

What clients say.

“Amar did an excellent job. He reviewed and finalized our AI & data engineering system. He quickly understood the existing architecture, fixed NLP and data pipeline issues, and improved overall system stability. Communication was clear, work was well-documented, and everything was delivered on time. We would definitely work with him again for future enhancements.”
Co-FounderBaynton Jesse
“Out of all of the app development teams that I contacted, this team understood my project the best. They created a great product, not breaking.”
FounderGiisty
“I had great experience with this Amar. He went beyond my requirements, super fast, skilled and highly recommended for other clients.”
Restaurant OwnerRara Restaurant
“He is professional. We have been working on this project for a few months and he has always been fantastic to work with. I look forward to future collaborations.”
Product OwnerHyves
“Very good communication, delivered on time. It was nice working with him!”
Lars SteinschulteRuhr-Universität Bochum
“We have been working on this project for a few months and he has always been fantastic to work with. Highly Recommended!”
OwnerMasal & TWOBIRDS Bridal

Process

How we'll work together.

A simple, transparent process — so you always know what's being built, why, and when.

Step 01
Discovery
A free call to understand your goals, constraints and timeline — with honest advice on what to build, and what not to.
Step 02
Architecture
A clear technical plan before any code: system design, stack, milestones and API contracts, sized to your budget.
Step 03
Build
Short iterations with working demos, code reviews and automated tests, so you see real progress early and often.
Step 04
Launch & scale
Production deployment with monitoring and auto-scaling, then documentation and handover so your team owns it.

About

The engineer behind the work.

AI Engineer & AI Architect

I'm an AI engineer and AI architect with 12+ years of experience taking AI from idea to production. I've led engineering teams at Giisty and Hyves.co in Dubai, and today I work with startups and enterprises worldwide as a freelance AI engineer and consultant.

My work spans generative and agentic AI, model training on custom datasets, computer vision, NLP and data science, image and video generation pipelines, and the MLOps and GPU infrastructure to run it reliably. I design RAG systems, build multi-agent workflows with LangGraph, CrewAI and MCP, and fine-tune models for real business use. I'm also a published researcher (PLOS ONE) with an MS in Computer Science earned on a full scholarship.

Location: Dubai, UAE · remote worldwide
Time zone: GST (UTC+4)
Email: dr.amar.sohail@gmail.com
Phone / WhatsApp: +92 304 9268347

Currently exploring

Agentic AI & multi-agent orchestration
Video generation models
Claude Code & MCP
GPU inference optimization

Highlights

AI at scale
AI content platform, 100 characters an hour
Built Dataism end to end: 30+ ComfyUI workflows for Flux LoRA training, SDXL image generation, voice cloning and video generation, running 48h+ continuously on GPUs.
View on GitHub
Published research
Peer-reviewed research
Published in PLOS ONE: formal modelling of IoT routing-protocol security with coloured Petri nets — a first-of-its-kind approach to verifying RPL vulnerabilities.
Read the paper
Scholarship
MS on full scholarship
Master's in Computer Science at COMSATS University on a merit-based full scholarship, with a focus on data mining, simulation and modelling, and cybersecurity.

Experience

12+ years of shipping.

Work

2024 – Present
Freelance AI Engineer & Consultant
Remote · clients worldwide
Design and deliver production AI systems for startups and enterprises — RAG architectures, multi-agent orchestration, LLM fine-tuning and AI workflow automation with n8n, MCP and Claude Code — from strategy through deployment.
- Python
- LangChain
- LangGraph
- CrewAI
- PyTorch
- MCP
- n8n
- AWS
- Kubernetes
- Terraform
2022 – 2024
Tech Lead
Hyves.co · UAE & France
Led the engineering team building a social commerce platform. Architected scalable backend services, owned deployment pipelines and technical decisions across the product lifecycle, and integrated LLMs and RAG for product recommendations and search.
- Node.js
- React
- PostgreSQL
- Docker
- AWS
- LangChain
- RAG
- n8n
2020 – 2022
Engineering Lead
Giisty · Dubai
Led cross-functional teams across backend, AI/ML and cloud. Architected microservices on Kubernetes (EKS), designed RAG pipelines with vector databases, shipped LLMs via SageMaker and MLflow, and set up GitOps CI/CD, Terraform and Prometheus/Grafana observability.
- Python
- PyTorch
- Kubernetes
- Terraform
- AWS
- PostgreSQL
- Redis
- Kafka
2018 – 2020
Backend Engineer
Digital Innovation (Pvt) Ltd · Lahore
Built REST APIs and event-driven microservices handling 100K+ daily requests. Implemented OAuth 2.0, role-based access control and encryption, and improved query performance by 40% through database redesign.
- Spring Boot
- Node.js
- NestJS
- PostgreSQL
- MySQL
- AWS
- Redis
- GraphQL
2016 – 2018
Software Engineer
University of Central Punjab · Lahore
Modernized full-stack web applications, migrating legacy PHP to Laravel MVC, and built REST APIs, automated tests and CI/CD for student platforms serving 10K+ users.
- PHP/Laravel
- JavaScript
- Node.js
- MySQL
- Linux
- Nginx

Education

2016 – 2019
MS Computer Science (Full Scholarship)
COMSATS University Islamabad, Lahore Campus
Merit-based full scholarship. Research published in PLOS ONE. Focus areas: cybersecurity, simulation & modelling, data mining.
2011 – 2015
BS (Hons) Software Engineering
Government College University Faisalabad
Software project management, programming, database systems, web engineering, data structures & algorithms, artificial intelligence.

Certifications

IBM RAG and Agentic AI Professional Certificate
IBM · Coursera · June 2025
Generative AI with Large Language Models
Coursera · March 2024
Google Cloud Professional Architect
Google Cloud · September 2023
AWS Certified Cloud Practitioner
Amazon Web Services · April 2018

Tech stack

Tools I use to deliver.

I pick the stack that fits your problem and your team — not the other way round. These are the tools I've shipped to production with.

Generative & agentic AI

OpenAI API
Claude
Gemini
LangChain
LlamaIndex
LangGraph
CrewAI
AutoGen
OpenAI Swarm
MCP
Claude Code
n8n
RAG
Pinecone
Vector databases

Model training

PyTorch
TensorFlow
Hugging Face
LoRA / QLoRA
scikit-learn
Datasets
Synthetic data

Image, video & voice

ComfyUI
Flux
SDXL
LoRA training
Video generation
Voice cloning
Whisper

Computer vision & NLP

OpenCV
Detection & segmentation
OCR
Transformers
Embeddings
NER

Data science

Python
pandas
SQL
ETL pipelines
PostgreSQL
MongoDB
Elasticsearch
Redis

MLOps & GPUs

Kubernetes
Docker
MLflow
SageMaker
FastAPI
VRAM optimization
Terraform
AWS
Azure
GCP
Prometheus
Grafana

Languages

Python
TypeScript
Go
Java
Rust
SQL

GitHub activity

@digitalinnovator

Repositories: 62+
Pull Requests: 280+
Contributions: 1200+

Blog

Latest insights.

Practical write-ups on production AI, agents, infrastructure and backend architecture.

All articles

FAQ

Questions, answered.

Something else on your mind? Ask me directly — I'm happy to talk through your project.

Book a free call WhatsApp me

What do you specialize in?

I'm Amar Sohail, a freelance AI engineer and AI architect with 12+ years of experience. I specialize in generative AI and RAG, agentic AI and multi-agent systems, model training and fine-tuning on custom datasets, computer vision, NLP, data science, image and video generation, and the MLOps and GPU infrastructure that runs AI in production.

Are you available for freelance or contract work?

Yes. I take on freelance AI engineering, consulting and contract work — from AI audits and prototypes to full builds and long-term team augmentation. Book a free 30-minute call, or hire me on Upwork.

How do engagements and pricing work?

I work on both project-based and hourly engagements. After a free discovery call, you get a clear scope and a custom quote based on your goals, timeline and budget.

Can you build a RAG pipeline or AI assistant for my business?

Yes. I design and deploy production RAG pipelines and AI assistants with LangChain, LlamaIndex and vector databases such as Pinecone, Weaviate and ChromaDB — including chunking strategy, embedding models, retrieval tuning and evaluation, so answers stay grounded in your data.

Can you build AI agents and multi-agent systems?

Yes. I build agentic AI with LangGraph, CrewAI, AutoGen and OpenAI Swarm, and agentic workflows with MCP (Model Context Protocol) and Claude Code — including tool use, persistent memory, guardrails and human-in-the-loop review.

Can you train or fine-tune models on our own data?

Yes. I prepare and curate datasets (including labelling and synthetic data), fine-tune LLMs and diffusion models with LoRA/QLoRA, train custom PyTorch and TensorFlow models on GPUs, and evaluate them against your real use cases before deployment.

Do you work on computer vision, NLP and image or video generation?

Yes. I build computer vision (detection, segmentation, OCR), NLP and speech (classification, entity extraction, summarization, Whisper speech-to-text), and generative media pipelines with ComfyUI, Flux and SDXL — including LoRA training for consistent characters, video generation and voice cloning.

How do you run AI models in production?

With MLOps: containerized model serving, GPU inference and VRAM optimization, CI/CD for models, monitoring and drift detection, and autoscaling on Kubernetes across AWS, Azure and GCP — with MLflow and SageMaker for the model lifecycle.

Where are you based, and do you work remotely?

I'm based in Dubai, UAE, and work remotely with clients worldwide, with projects delivered for teams across the US, Europe and the Middle East.

Contact

Let's build something great.

Tell me about your project — the goal, the timeline and where you're stuck. Or skip the form and book a free 30-minute call.

Tell me about your project

Prefer to talk? Book a free 30-min call

Walk me through your idea or problem and leave with honest, practical next steps — no commitment.

Pick a time

I build AI systems that ship to production.

Trusted by teams at

What I can build for you.

Generative AI & LLM Apps

Agentic AI & Automation

AI Architecture & Strategy

Model Training & Fine-tuning

Image & Video Generation

Computer Vision

NLP & Speech

Data Science & Datasets

MLOps & GPU Infrastructure

Selected work.

MakeInfluencer.ai

Magical Children's Book

JoyPix.ai

HappyHorse

Dataism

Giisty

HYVES

KeyPact PQ-Mail Proxy

Gold Bank London

Just Eat

Asurint

Salt'n Pepper

TWOBIRDS Bridal

QKSMS

Wispy

What clients say.

How we'll work together.

Discovery

Architecture

Build

Launch & scale

The engineer behind the work.

Amar Sohail

Highlights

AI content platform, 100 characters an hour

Peer-reviewed research

MS on full scholarship

12+ years of shipping.

Work

Freelance AI Engineer & Consultant

Tech Lead

Engineering Lead

Backend Engineer

Software Engineer