Khadim HussainAI Engineer & Consultantfor Computer Vision & LLMs
I build defect detection and visual inspection systems, speed up model inference with TensorRT and ONNX, and deploy fine-tuned LLMs and RAG pipelines to production.
Khadim Hussain
AI Engineer & Consultant
Transforming Businesses Through AI Innovation
Khadim Hussain is a results-driven AI engineer and consultant with expertise in building production-ready intelligent systems. Currently working at elunic AG (Munich), he specializes in Computer Vision, LLM Integration (ChatGPT, Claude, GPT-5), AI Agent Development with LangChain, and RAG Systems.
With a proven track record of achieving 99.999% model accuracy and 70% inference speed improvements using PyTorch, TensorFlow, TensorRT, and ONNX, Khadim delivers Generative AI, Fine-Tuning (LoRA), and Multi-Agent Systems deployed via Docker, Kubernetes, and MLOps pipelines.
From startups to enterprises, Khadim has delivered AI solutions including Object Detection (YOLO11), Image Segmentation, RAG Chatbots, Voice AI, and Enterprise Automation systems—always focusing on scalable, production-ready deployments with measurable business impact.
Core Competencies
Experience Timeline
Pushing the boundaries of AI/ML across industries with cutting-edge solutions
elunic AG
Munich, Germany
Senior Machine Learning Engineer
speed
70% faster
role
Team Lead
scope
Architecture
Architected and built internal inference engines and Python tooling used across the ML team, optimizing model serving through TensorRT, ONNX, and Triton, resulting in a 70% improvement in inference speed
Led deep learning research initiatives and developed innovative engines for production deployment across multiple providers
Mentored and onboarded new team members, establishing training workflows and best practices for the ML team
Data Scientist
accuracy
99.999%
improvement
29.999%
models
Production
Developed and fine-tuned segmentation and detection models using PyTorch and TensorFlow, deploying them with Docker and Kubernetes in an MLOps environment
Achieved an accuracy rate of 99.999%, significantly improving company model performance from 70% to 99.999%
Led data annotation efforts and managed a team to ensure high-quality annotated datasets for model training
Freelance (Upwork)
Toronto, Canada
datasets
Large-scale
efficiency
Optimized
models
Production
Collected and annotated large-scale text datasets for LLM projects
Built text classification models for Q&A systems
Developed efficient data collection and annotation Pipelines for streamlined workflow
Freelance (Upwork)
New York, USA
speed
90% faster
scope
CV & LLM
deployment
Production
Built End-to-End TensorFlow models for Computer Vision and LLM applications
Achieved 90% faster predictions in inference pipelines
Optimized training pipelines for peak performance
Ready to leverage this expertise for your next AI project?
Let's Build TogetherServices & Expertise
Hire an AI engineer and consultant for computer vision, LLMs and AI agents, from the first assessment to a system running in production
Computer Vision Consulting
Defect detection and visual inspection for production lines, from checking your images and labels to a model tested on your line's hardware.
Computer Vision Consulting detailsTensorRT Optimization Consulting
Trained models made faster on the GPUs you already have with TensorRT, ONNX Runtime and Triton, each variant measured against the original.
TensorRT Optimization Consulting detailsLLM Fine-Tuning & Consulting
Open LLMs fine-tuned for a defined task, evaluated against a baseline, and delivered as a model you can run with Ollama, vLLM or Transformers.
LLM Fine-Tuning & Consulting detailsAI Agent Consulting
Agentic workflows, RAG, chatbots, MCP servers and AI automation with n8n, built to hold up in production rather than only in a demo.
AI Agent Consulting detailsFeatured Projects
Explore my portfolio of AI-powered solutions that have transformed businesses across industries
AI Memory: multimodal semantic search for AI agents
A self-hosted AI memory service for text, images, audio and video, using CLIP, ImageBind, Whisper, Qdrant and a PostgreSQL graph.
AI-Powered Micrograph Analysis
AI solution for automated micrograph (Schliffbild) analysis in welding and material inspection. Uses deep learning for segmentation, keypoint detection, and classification to identify weld types, detect cracks/pores, and calculate precise measurements. Reduced inspection time by 80% with human-level accuracy (0.0001mm error margin).
PackTrace™ - Advanced Package Detection System
State-of-the-art industrial computer vision system for high-speed conveyor belt operations. Delivers 99.7% accuracy in real-time object detection, quality control, and defect identification at 120+ FPS with sub-10ms latency.
Kiffusion and SpaceVision: a FLUX LoRA training dataset and space LoRAs
How I built a FLUX LoRA training dataset of 1,964 space images, captioned with JoyCaption, and trained LoRAs for FLUX.2-klein (9B) and Kolors.
STEM reasoning LLMs: fine-tuning Qwen3-14B and gpt-oss-20b with Unsloth
How I fine-tuned Qwen3-14B and gpt-oss-20b with Unsloth and QLoRA for STEM Q&A, then shipped LoRA, merged and GGUF builds for Ollama and llama.cpp.
Latest Writing
Measured tests and how-tos from real computer vision, LLM and AI agent work
AI Agents
Claude Code on Windows: install it and use it with no coding
Install Claude Code on Windows with one PowerShell command, fix the 'claude is not recognized' error, sign in, and run two tasks with no coding.
· 10 min read
AI Agents
Claude Code skills: how they work, and when Claude uses one by itself
How Claude Code skills work, where SKILL.md goes and why one won't trigger, with a 108-run test of skill names, descriptions and manual-only skills.
· 9 min read
LLM Fine-tuning
Fine-tuning gpt-oss-20b with Unsloth and running it in Ollama
How to fine-tune gpt-oss-20b with Unsloth: QLoRA settings, VRAM, GGUF export and a working Ollama Modelfile with the Harmony template and stop token.
· 8 min read
Client Testimonials
Hear what my clients have to say about their experience working with me
“Working with Khadim has been an absolute pleasure. He is a skilled and proactive engineer who consistently delivers top tier results. Over the past year he has played a crucial role in successfully executing multiple complex projects with high code quality and clear communication. He takes strong ownership of his work, solves difficult challenges independently and always meets deadlines. I highly recommend Khadim to anyone looking for an elite developer. He is also a strong Agentic developer that can automate anything.”
Nodirbek H.
Full Stack Developer · Aug 2026
“Khadim is a professional and pays a lot of attention to the cases. Happy with the consultation.”
Adrian L.
30 minute consultation · Aug 2025
“Khadim was able to deliver the project on time with flying colors”
Amber Z.
Text analytics using Python · Oct 2022
“Khadim was successfully able to complete both my tasks on Text classification within the time agreed on. He has been very flexible and has a great work ethic. Would definitely reach out to him again for any of my future tasks and recommend him to anyone looking for a great end product!”
Karthikeya Reddy G.
Text Classification 2 · Jun 2022
“Khadim was able to deliver the work I needed extremely fast, within a day, and did excellent work with my machine learning work. Definitely hire him he is an expert in this field. He is very fast to reply to messages and makes fast revisions I am very impressed at how professional he is.”
John G.
Code linear regression · May 2022
5.0/5 Upwork rating · 100% Job Success · Top Rated Plus · $250K+ earned
Start Your AI Journey