Frame the problem
Start with the decision a person needs to make, the evidence available, and what failure would cost.
Product thinking · Research · System designIndependent engineer · Islamabad, PakistanOpen to collaborations
Portfolio / 2026Building
Building
intelligence
into products.
I'm Muhammad Bilal Tahir, a full stack AI engineer. I turn ambitious ideas into useful software, from agentic workflows and computer vision to complete web products.
Scroll to discover01 / 06
A demo is easy to admire. The real test starts when the answer must be correct, useful, and still there when something fails.
02 / The evidenceFrom question
From question
to working system.
Follow six real problems from first question to engineering decision. Choose a case to explore the build.
Choose a case, or use the arrow keys
01 / 06
03 / The methodOne engineer.
One engineer.
Across the stack.
I work backward from a real decision: what a person needs to know, what the system can prove, and how it behaves when the ideal path breaks. Then I build across the stack to make that decision easier.
Ground the intelligence
Choose the right model and data path, measure uncertainty, and make every answer or prediction traceable.
LangGraph · RAG · PyTorch · YOLO · EvaluationBuild the experience
Connect the interface, APIs, data, and authentication so the clever part becomes a useful product.
Next.js · React · TypeScript · FastAPI · PostgreSQLKeep it dependable
Plan fallback paths, ship through CI/CD, and make the system observable and operable after launch.
Docker · Kubernetes · Redis · Linux · Azure03 / From idea to productYou bring the
You bring the
business idea.
I shape it, build the full product, and get it ready to launch—from the first user flow to the working system behind it.
Let's build itIdeaProductLaunch
04 / The practiceExperience that
Experience that
ships.
Research depth, production constraints, and client delivery shape the way I build.
Full Stack Developer
Markhor SystemsBuilding client products end to end, including a medical LMS with Zoom automation, attendance, recording uploads, and live community chat.
Computer Vision Research Intern
NCRA UAV Dependability LabBuilt RGB and thermal datasets for 4 aerial-object classes; benchmarked 5 detectors and helped develop SkyEye’s confidence-gated YOLO → VLM pipeline. Its RGB model reached 94.11% mAP50 in evaluation.
AI Engineer Intern
Prompt To Flow · NICBuilt a FastAPI workflow engine that converts natural language into executable graphs, with provider fallback, tenant credit gating, and key rotation.
Software Engineer Intern
RainMakerzContributed to a serverless Python and FastAPI backend on Azure Functions in a 19-agent document processing system.
05 / The foundationStill learning.
Still learning.
Always building.
BS Software Engineering at FAST-NUCES, Islamabad, expected 2028. CGPA 3.67/4.00 with four Dean's List honors. Four-time lab demonstrator across Object-Oriented Programming and Data Structures.
See all credentials on LinkedInSelected certificates
- Build AI Agents using MCPIBM · Aug 2026
- Neural Networks and Deep LearningDeepLearning.AI · Aug 2026
- Retrieval Augmented Generation (RAG)DeepLearning.AI · Jul 2026
- Develop Generative AI Applications: Get StartedIBM · Jun 2026
- CS50's Introduction to Artificial Intelligence with PythonHarvard · Mar 2026
- RAG with Vertex AIGDG on Campus PIEAS · Jan 2026
- CS50's Introduction to Computer ScienceHarvard · Aug 2025