Service
AI, LLMs & data automation.
Local LLMs, scraping pipelines and custom data tools that turn websites, PDFs and images into clean datasets and live dashboards.
01 · Overview
How I help.
At NUST SEECS I pooled an RTX 5080, an RTX 4080, an A100 and a 128 GB GPU into one cluster so AI engineers can run Llama and other models locally, and I deploy, fine-tune and test LLMs on it. For S2Cool I scraped 15 years of weather data from across Pakistan, turned it into formulas and published it as a live map. I also build extraction tools for publishers and install LLM apps on servers.
- Python
- Selenium
- Local LLMs
- NVIDIA RTX
- OCR
- Next.js
- Vercel
02 · What's included
Everything,
end to end.
Local LLM deployment
Models installed, fine-tuned and tested on multi-GPU rigs.
LLM apps on servers
Self-hosted AI tools installed and secured on client servers.
LLM-powered scraping
Scrapers that adapt to each site with LLMs, plus per-publisher scripts for protected sites.
PDF & image extraction
Emails, author names, titles and more pulled from PDFs, plus OCR for scans.
Formulas & analysis
Means, deviations, anomalies and derived metrics.
Dashboards & maps
Data published as live, interactive web apps.
Proof · related case studies
FAQ
Questions, answered.
Can you run LLMs locally instead of using paid APIs?
Yes. I deploy and fine-tune local LLMs on multi-GPU NVIDIA RTX machines, which keeps data private and costs predictable.
What kind of data can you extract?
Websites (including JavaScript-heavy pages), PDFs and scanned images. For S2Cool I extracted 15 years of Pakistani weather data.
Can you turn the data into a dashboard?
Yes. S2Cool publishes its data as a live Next.js map with parameter layers, deployed on Vercel.
From the articles
The full
write-up.
LLM-powered scraping: why smart scrapers beat brittle selectors
Scrapers break when sites change. How I use LLMs to read each layout, script each publisher and pull emails, authors and titles from PDFs.
Pooling GPUs for local LLMs: lessons from a university lab
How we pooled an RTX 5080, RTX 4080, A100 and a 128 GB GPU into one shared cluster for Llama and other local LLMs at NUST, and what to plan for.