Browse categories
Explore
Fiverr Pro
English
$
USD
I will fine tune llm gpt or bert, train custom ai model on your dataset
Abbas Mustafa
About this gig
You want a model that actually understands your domain, not a generic ChatGPT wrapper.
I fine tune open source LLMs (Llama, Mistral, Qwen, DeepSeek) and GPT models using LoRA, QLoRA and PEFT on your custom data: chat assistants, classification, summarization, data extraction, or deep domain expertise (legal, medical, finance, support).
What you get:
Fine tuned model in Hugging Face format, ready to deploy
Clean training and inference code
Evaluation report: base model vs fine tuned
Reproducible pipeline (data prep, train, evaluate)
Optional deployment as an API or quantized GGUF
Why me:
ML engineer, 5+ years building production AI systems
Small dataset? I handle it with augmentation and synthetic data
Documented handover and clear communication
I also fine tune classic transformers (BERT, T5) when they fit the task better and run cheaper.
Message me with your task and a data sample. I will tell you honestly if fine tuning is right, or if RAG or prompting solves it cheaper.
Get to know Abbas Mustafa
Abbas Mustafa
AI Engineer for LLM, RAG, Chatbot and ML Pipeline Solutions
- FromFrance
- Member sinceDec 2018
- Avg. response time1 hour
- Last delivery1 year
Languages
English
I'm an AI Systems Engineer with 5+ years of experience building production-grade AI solutions — RAG pipelines, fine-tuned LLMs (LoRA/PEFT), multi-agent systems, and computer vision applications. I've deployed ML systems for medical imaging, trading, and security, and built full-stack platforms serving 10,000+ users with FastAPI, Docker, and Kubernetes. Currently pursuing an MSc in Cybersecurity in France, I bring a security-first mindset to every build. Whether you need a chatbot, an automation agent, or a complete ML pipeline, I deliver clean, scalable, well-documented work.
My Portfolio
Other AI Development Services I Offer
FAQ
What do I need to provide to get started?
Your dataset (CSV, JSON, or text files), a short description of the task, and the metric you care about. If your data is unlabeled or you are unsure about the format, message me first and I will guide you.
Can you work with sensitive or private data?
Yes. Your data stays confidential and is only used for your project. On request I can train on your own infrastructure (your cloud or server) so the data never leaves your environment, and I am happy to sign an NDA.
