Hi, I'm Muhammad Talha — most people call me TJ.

I build things for the web, then go read about hobbits, benders, and failed states.

Full-Stack Developer in Dallas–Fort Worth, TX. M.S. Computer Science, UT Dallas.

About

What I actually do

I'm a full-stack developer currently building internal tools that keep a construction company's field crews and back office looking at the same numbers. Before that, I was on the frontend of a SaaS platform in San Francisco, shipping features people actually opened every day.

I like software that removes friction from something a human used to do by hand — a spreadsheet, a paper log, a Slack thread nobody trusts. Most of my work lives at that intersection: React and Next.js on the front, Node and Postgres (or whatever the job actually needs) underneath.

In January 2026 I started an M.S. in Computer Science at UT Dallas. Alongside that, I'm doing graduate security research measuring whether AI-generated code patches are actually as safe as the human-written ones they're replacing — turns out that's a harder question than it sounds.

Outside of code, I read a lot of books about how nations convince themselves they're nations, and I have never once regretted a Lord of the Rings extended-edition marathon.

Education

M.S. Computer Science
University of Texas at Dallas · Jan 2026 – Aug 2028
B.S. Computer Science
Habib University · Jun 2020 – Jun 2024

Stack

PythonJavaScriptTypeScriptSQLReactNext.jsTailwindCSSReduxReact RouterNode.jsExpressFastAPIRedisMySQL

Experience

Where I've worked

Full-Stack Developer · Jacqueline Construction

Wylie, TX

Led development of end-to-end job and time-tracking systems used by 50+ field and office staff, reducing manual reporting effort by 70%.
Built scalable APIs and frontend workflows using React, Redux, Node.js, and MySQL to enable real-time job status, labor cost tracking, and operational visibility — cutting the daily reporting workload by 250+ hours annually.
Worked directly with operations and management to translate loosely defined business needs into shipped features on biweekly timelines.

Oct 2024 – Present

Full-Stack Developer (Frontend-Focused) · QLU.AI

San Francisco, CA

Delivered multiple end-to-end features for a production SaaS platform using React, Next.js, TailwindCSS, and Node.js, supporting 300+ daily active users.
Implemented structured frontend state management with Redux and React Router, reducing UI regressions by 30%.
Built full-stack attachment functionality spanning React UI and AWS S3/EC2, using pre-signed URLs for secure uploads that never route files through the app server.
Built a reverse-indexing and tokenization system on Apache Cassandra to enable search over deterministically encrypted messages — fast lookup, automatic index updates on ingestion, zero plaintext storage on client or server.

Jun 2024 – Oct 2024

Projects

What I've been building

To Vibe or Not to Vibe

May 2026 – Current

Graduate security research with a UT Dallas professor, targeting publication: evaluating AI-generated patch quality against human-written fixes using CodeQL static analysis across the SWE-bench Pro benchmark.

Built an end-to-end analysis pipeline that clones 11 open-source repos (JS, Python, Go), applies patch variants (control / golden / AI) via git worktrees, and runs CodeQL security-extended query suites automatically.
Engineered a SARIF diffing engine to compare findings before and after patching at the (ruleId, file, line) level, quantifying vulnerabilities fixed vs. regressions introduced per patch.
Leveraged HPC resources to scale CodeQL database creation and query execution across the full SWE-bench Pro test split, handling large multi-language codebases at volume.

PythonCodeQLSecurityHPCResearchRead the full paper

Research