What I build
Retrieval and extraction systems that measure themselves: evaluation sets, confidence scores, automated tests and written failure modes.
BS Computer Science · Iqra University Islamabad
Trustworthy AI for Urdu and Pakistani public information.
I am Salik Hussain, a first-year computer science student building PAKGOV-RAG, a bilingual Urdu-English retrieval system for Pakistani government documents. I publish measured results, tests and failure cases, and I am working toward an MS in AI abroad.
Rawalpindi, Pakistan · Open to research internships, remote Python and AI work.
- Urdu script
- پاسپورٹ کی فیس کتنی ہے؟
- Roman Urdu
- Passport ki fee kitni hai?
- Urdu-English mixed
- پاسپورٹ کی fee کتنی ہے؟
Illustrative inputs only, not model output. The open question: does a system find and cite the same correct source for all three?
- LeetCode problems solved
- 795
- Longest daily streak
- 51 days
- HackerRank certificates
- 10
- Harvard CS50 certificates
- 5
Read from my public profiles in October 2026. Peer-reviewed publications: none yet. Everything on this page is finished, in progress, or marked Planned.
If you have 30 seconds
Best proof
PAKGOV-RAG compares BM25, dense and hybrid retrieval on 50 English, Urdu and mixed questions. Hybrid reached R@5 of 0.74.
Read the case studyWhat I am looking for
Research internships and supervision in Urdu NLP and RAG, remote Python or AI work, and a fully funded MS in AI (planned entry 2030).
Get in touchExplore
Research
The question I am working on, five planned research projects, and a four-year plan with every item labelled by status.
Open researchProjects
PAKGOV-RAG in full, with measured results and limits, plus five other repositories.
Open projectsRecord
LeetCode and HackerRank figures, 21 certificates with verification where available, education and experience.
Open record