About
AI engineer at Symufolk. I build the pipeline behind Puffo AI.
I'm an AI engineer at Symufolk, where I work on the Puffo AI platform. I joined as an intern in April 2026 and was promoted to Associate AI Engineer in July. The system I spend most time on is a nine-agent pipeline built as a forward-only state machine: human approval gates between phases, an append-only audit log, folder-scoped read and write permissions per agent, sandboxed execution for the agent that runs code, and an untrusted-input policy for the student-facing agent. Heavy reasoning models go only to the agents making judgment calls; the high-volume agent runs on a cheap model grounded by a RAG pack.
Seven of the 321 commits on Provider Research Factory are mine. They are the founding architecture and the first hand-verified end-to-end run. The other 314 were made by an orchestrator agent I defined, provisioned and gated. The build streams about 9.8 million NPPES registry rows, halts on any of 11 assertions, runs a compliance gate written as non-bypassable code, and deliberately ships no send path.
I finish my BS in Data Science at the University of the Punjab in October 2027. I'm also an AI/ML trainee at NETSOL Technologies through December 2026. I'm remote from Lahore, 1:00–10:00 PM PKT, a full working day's overlap with London. I work best where the automation has to be defensible to someone outside the team before it ships.
Chess and card games otherwise.
HOW I WORK
Four rules the pipelines are built on.
- 01
Orchestrate deterministically.
Use models only where judgment is actually required. The control flow, the state transitions and the stopping conditions are code.
- 02
Gates are code, not policy documents.
If it can be skipped under deadline pressure, it was never a gate.
- 03
An unexercised guard is not a guard.
I run every guard at least once against a case it should catch.
- 04
Label the gap, let it surface.
When a step can't be completed, I say so rather than dropping it silently.
Credentials
Every line links to where you can check it.
Education
BS Data Science
University of the Punjab · Lahore
Expected October 2027
Certifications · 9
Exact titles as printed. Verification links, not PDFs.
Supervised Machine Learning: Regression and Classification
DeepLearning.AI and Stanford University via Coursera
25 May 2025VerifySQL (Advanced)
HackerRank
13 Apr 2025VerifyLinear Algebra for Machine Learning and Data Science
DeepLearning.AI via Coursera
19 Jan 2025Verify
Show 6 more (foundation courses)Show fewer
Extract, Transform and Load Data in Power BI
Microsoft via Coursera
28 Jan 2025VerifyHarnessing the Power of Data with Power BI
Microsoft via Coursera
24 Jan 2025VerifyPreparing Data for Analysis with Microsoft Excel
Microsoft via Coursera
16 Jan 2025VerifyTools for Data Science
IBM via Coursera
25 Dec 2024VerifyWhat is Data Science?
IBM via Coursera
22 Dec 2024VerifySQL (Basic)
HackerRank
13 Apr 2025Verify
Achievements
- Winner, Coding HackathonKennard College University
- Winner, Coding HackathonBahria University
- 350+ problems solvedLeetCodeVerify
- 150+ problems solvedCodeforces
- Campus AmbassadorArtificial Intelligence Community of Pakistan