Browse categories
Explore
Fiverr Pro
English
$
USD
I will fine tune, quantize, and deploy open source ai models
Aaliyan Khan
About this gig
Want your own custom AI model, not just a prompt wrapped around GPT 5, but one trained on your data and optimized to run efficiently?
I fine-tune, quantize, and deploy open-source AI models so you fully own the result: no per-token fees, no third-party data sharing, behavior built for your exact use case.
WHAT I DO:
- Fine-tuning: LoRA, QLoRA, full fine-tuning on Mistral, Qwen, Deepseek, GLM
- Quantization: GGUF, GPTQ, AWQ for faster, cheaper inference
- Deployment: model + setup docs for your server, cloud, or device; direct setup with access, or live guidance
- Image models: Stable Diffusion fine-tuning available
80+ five-star Fiverr reviews, [91] completed projects (software/game dev background). Hands-on model training, not API reselling.
WHY OPEN SOURCE:
- Full data privacy if self-hosted
- No per-token costs long-term
- Domain-specific behavior, no lock-in
I'll be honest if GPT 5 with good prompting is actually the better fit instead.
WHAT YOU GET:
- Model trained/optimized for your use case
- Benchmarks vs. base model
- Docs to run, retrain, adjust
- Setup support per package tier
Unsure which package fits? Message me first, I'll give an honest read.
Get to know Aaliyan Khan
Aaliyan Khan
Full Time Minecraft Mod Dev
- FromPakistan
- Member sinceNov 2022
- Avg. response time1 hour
- Last delivery6 days
Languages
Urdu, Punjabi, English
Hey everyone! I'm AbdulHanan, also known as Aaliyan Khan, and I'm passionate about game modding, especially for Minecraft, I've 6 years of Minecraft modding experience since 2020. Now Working Full Time as Minecraft Mod Dev on Fiverr. I've worked on tons of awesome modded servers and I'm excited to share my knowledge. If you need help with your mods, videos, or just brainstorming ideas, don't hesitate to ask!
My Portfolio
Other AI Development Services I Offer
FAQ
Will you set this up directly on my device or server?
Depends on the package and access you can provide. If you give me SSH/remote access to a server or cloud instance, I can configure it there directly. For a personal device (laptop, phone), I deliver the model plus setup instructions, and Standard/Premium include support if you get stuck — Premium
What's the difference between fine-tuning and quantization?
Fine-tuning changes what the model knows or how it behaves (trained on your data). Quantization changes how efficiently the model runs (smaller, faster, less memory) without retraining its knowledge. Many projects benefit from both — I'll tell you which one(s) you actually need.
Do I need my own GPU/server?
Not necessarily. I can train using cloud GPU rental and quantize the result for lightweight deployment, including on regular CPUs via llama.cpp for smaller models. You'll need somewhere to eventually run the finished model, but it doesn't have to be powerful hardware if we quantize appropriately.
Can you fine-tune image models too, not just text?
Yes, Any Image model fine-tuning (LoRA style) is available as an add-on. Let me know in the requirements if this applies to your project. Krea2 etc with permissive license.
Is fine-tuning actually necessary for my use case, or would GPT-4/Claude with good prompting be enough?
Often prompting alone gets you 80% of the way there for less money. I'll tell you honestly if that's true for your case before you pay for a bigger package.
How much data do I need for decent fine-tuning results?
Depends on the task, but a few hundred well-structured examples can already show meaningful improvement for narrow tasks. I'll assess your data during the requirements stage.
Will you use my data to train anything other than my own model?
No, your data and resulting model are yours. Nothing is reused for other clients or shared elsewhere.
