DemoDemo

AI / ML ENGINEER • OPEN SOURCE • RESEARCH

Building practical AI systems for the real world.

I design, build, and deploy machine learning systems that are efficient, reliable, and useful. My work spans model serving, RAG, developer tooling, and applied research.

View my work Download résumé
  • Islamabad, Pakistan
  • BS Computer & Information Sciences
  • Open to opportunities
Muhammad Ali Nasir

FEATURED PROJECT

Every card opens a detailed case study

01 / DEVELOPER TOOLS

CodeGraph CLI

A codebase intelligence tool that turns source files into a navigable graph for faster exploration, retrieval, and agent workflows.

40+
languages
AST
code parsing
Graph
aware retrieval
CLI
open-source tool
Open full case study GitHub ↗
CodeGraph CLI codebase explorer
Source-code evidence from the CodeGraph CLI explorer.
02 / KNOWLEDGE SYSTEMS↗

Research Council

A multi-agent research assistant that uses a knowledge graph to retrieve and synthesize scientific papers.

Open full case study
Research Council interface
  • 4-agent workflow
  • Hybrid RAG
  • Neo4j
03 / MODEL INFRASTRUCTURE↗

LocalForge

A self-hosted AI control plane that makes multiple local models feel like one dependable endpoint.

Open full case study
LocalForge control plane dashboard
  • <5ms routing
  • Local-first
  • 1 endpoint

RESEARCH & OPEN SOURCE

View experience

Research and open source projects

CLIMATE ML

S2Cool

Solar forecasting and cooling demand prediction for practical energy decisions.

S2Cool solar and cooling intelligence interface
Open case study →485K+ records

AGENTIC WORKFLOW

ProposalBot

Slack-native multi-agent workflow for proposal drafting.

ProposalBot chat workflow interface
Open case study →Thread-native

OPTIMIZATION

FuelRoute Planner

Optimal fuel-stop planning across long-distance routes.

FuelRoute Planner route dashboard

NLP RESEARCH

LightUHope

Compact transformer for Urdu hope-speech classification.

LightUHope research model results
Under review · Springer0.92 Macro F13.2M Parameters

ENGINEERING PHILOSOPHY

From research to real-world systems.

I pair thoughtful research with practical engineering to create technology that is useful, explainable, and grounded in real constraints.

More about me
  1. 01Understand

    Study the problem and real constraints.

  2. 02Build

    Design and implement practical systems.

  3. 03Evaluate

    Measure with real data and improve.

  4. 04Deploy

    Ship, document, and create impact.

EXPERIENCE

Download résumé

A practice shaped by applied ML, research, and open source.

Associate AI Engineer, DevRolin Part-time

Fine-tuned and deployed transformer models for production feature pipelines, RAG systems, and FastAPI services.

Freelance ML Research Engineer, S2Cool

Built solar irradiance prediction and cooling optimization models across 485,000+ climate records.

First Author, LightUHope

Developed a lightweight Urdu hope-speech classifier with 0.92 macro F1 and 3.2M parameters.

Open-Source Maintainer, CodeGraph CLI

Created a multi-repo code intelligence tool with AST parsing and graph-aware retrieval.

ABOUT

I’m Muhammad Ali Nasir.

I’m a machine learning engineer focused on building practical systems where research meets real-world impact. I enjoy turning complex problems into simple, reliable tools. My work ranges from models and RAG pipelines to developer workflows.

Let’s work together
LocalForge memory and routing evidence
More useful systems are built around people.

GET IN TOUCH

Have an AI or ML project in mind?

Open to discussions around machine learning engineering, LLM application architecture, and practical AI systems.

Emailmuhammadalinasir00786@gmail.com
LinkedIn GitHub