Industrial software & applied AI

I build software for

AI & Full-Stack Software Engineer at Lean Automation. I own the architecture for an industrial IoT platform that runs inside air-gapped customer infrastructure in the oil & gas sector, and I lead a team of three. Event-driven microservices in ASP.NET Core and Node.js, sensor telemetry over MQTT and Kafka, integration with plant historians and field instrumentation (AVEVA PI, PI Vision, control loops), and applied AI where it earns its place.

terminal://muhammad-usama

  ██╗   ██╗██╗      █████╗ ██████╗ 
  ██║   ██║██║     ██╔══██╗██╔══██╗
  ██║   ██║██║     ███████║██████╔╝
  ██║   ██║██║     ██╔══██║██╔══██╗
  ╚██████╔╝███████╗██║  ██║██████╔╝
   ╚═════╝ ╚══════╝╚═╝  ╚═╝╚═════╝
     I N D U S T R I A L · S W     
[00]sysindustrial platform · air-gapped deployment
[01]netmqtt · kafka · event-driven ingestion
[02]ot aveva pi · pi vision · instrumentation · control loops
[03]apiasp.net core · node.js · nestjs microservices
[04]ai on-prem rag · fine-tuned llm · anomaly detection
[05]ui react · next.js · real-time operator dashboards
[06]opsdocker · kubernetes · offline registry
[07]ok shipping
$deploy --target production --air-gappedlive
air-gapped

Timeline

Professional Journey

Experience across AI engineering, full-stack product development, and applied medical AI research.

AI & Full-Stack Software Engineer

Feb 2025 – Present
Lean AutomationPakistanFull-time

Own the architecture for a production industrial IoT platform delivered into air-gapped customer infrastructure in the oil & gas sector, and lead a team of three engineers.

  • Own platform architecture and lead a team of three engineers
  • Decomposed a monolith into independently deployable ASP.NET Core, Node.js and NestJS services
  • Designed the event-driven data layer over Kafka and MQTT for field sensor telemetry
  • Integrated plant historians (AVEVA PI, PI Vision) and field instrumentation, working across control loops with instrumentation and loop-performance engineers
  • Built on-premise RAG pipelines and a fine-tuned LLM engineering co-pilot — no external model APIs
  • Delivered the full stack into air-gapped environments with Docker, Kubernetes and an offline registry
  • Designed operator interfaces aligned to existing HMI conventions
ASP.NET CoreC#Node.jsNestJSNext.jsReactTypeScriptKafkaMQTTPostgreSQLDockerKubernetes

Software Engineer — AI/ML

Feb 2024 – Feb 2025
Lean AutomationPakistanPart-time

Built AI/ML features and full-stack functionality across the platform — model development, data pipelines, and the services around them.

  • Developed ML models and data pipelines feeding production features
  • Built and shipped full-stack functionality across React/Next.js and Node.js services
  • Implemented CI/CD with GitHub Actions and Docker for reliable releases
  • Built real-time dashboards for KPI and anomaly monitoring
ReactNext.jsNode.jsNestJSPythonPyTorchFastAPIDocker

Associate Software Engineer

Jan 2023 – Jan 2024
Lean AutomationPakistanPart-time

Delivered production web platforms and automation pipelines across analytics, APIs, CI/CD, and deployment workflows.

  • Delivered production full-stack applications
  • Implemented GitHub Actions + Docker CI/CD for zero-downtime deployments
  • Automated reporting via SQL procedures and APIs
  • Built real-time dashboards for KPI and anomaly monitoring
ReactNext.jsNode.jsNestJSSQLDockerGitHub Actions

Research Assistant

Mar 2024 – Feb 2025
COMSATS University Islamabad (Abbottabad Campus)Abbottabad, PakistanPart-time

Led thesis research on medical AI for brain tumor analysis, focusing on model design, dataset engineering, and publication-ready outputs.

  • Developed a Dual-Head Neural Network (DHNN) for tumor type + grade prediction
  • Built end-to-end MRI preprocessing pipelines with OpenCV and Pandas
  • Achieved 93.2% classification accuracy with EfficientNetB0 backbone
  • Co-authored and submitted work to Elsevier Results in Engineering
PythonTensorFlowKerasOpenCVPandasMedical Imaging

Bachelor of Science in Computer Science (BSCS)

Mar 2021 – Jan 2025
COMSATS University Islamabad — Abbottabad CampusPakistanDegree

CGPA: 3.56 / 4.0. Focused on AI/ML systems, software engineering, and data-driven problem solving.

  • Thesis: Multi-Class Brain Tumor Classification and Grade Estimation using DHNN
  • Coursework: AI, ML, DSA, DBMS, Operating Systems, Distributed Systems
  • Built research and production projects spanning healthcare and industrial AI
AIMachine LearningComputer VisionDistributed SystemsOOPDatabases
Journey continues...

Portfolio

Featured Projects

Production industrial software and on-premise AI systems, alongside applied research. Built to run in real, constrained environments.

Featured
2025 – Present

Industrial IoT Platform — Oil & Gas

Distributed platform ingesting live field-instrumentation telemetry through event-driven microservices, integrated with plant historians (AVEVA PI, PI Vision). Decomposed from a monolith into independently deployable ASP.NET Core, Node.js and NestJS services and delivered into air-gapped customer infrastructure.

Air-gapped deploymentEvent-drivenIn production
ASP.NET CoreNode.jsNestJSKafkaMQTTAVEVA PI
Featured
2024 – 2025

Industrial AI Co-Pilot — On-Premise RAG & Fine-Tuned LLM

Fine-tuned LLaMA 3.2 with a retrieval pipeline so answers cite proprietary sources rather than model memory. Multi-step tool-calling via LangGraph, served through FastAPI. Fully on-premise — no external model APIs.

On-premiseRetrieval-groundedFine-tuned
LLaMA 3.2RAGLangGraphFastAPIPyTorch
Featured
2024

VAE-LSTM Fault Detection

Unsupervised anomaly detection for industrial equipment where labelled failure data is scarce. VAE + LSTM over multivariate sensor streams with calibrated reconstruction-error thresholds, served via FastAPI with real-time dashboards.

Early fault detectionMultivariate sensor streams
PyTorchVAELSTMFastAPIIndustrial AI
Featured
Research
2024

Brain Tumor Classification — Dual-Head Neural Network

Dual-Head Neural Network on an EfficientNetB0 backbone for simultaneous multi-class tumour classification and grade estimation from MRI (4.0M parameters, ~15ms inference). Manuscript under review at Elsevier Results in Engineering.

93.2% accuracy15ms inferenceManuscript under review
TensorFlowKerasMedical AIComputer VisionResearch
Featured
2025

Encrypted Traffic Fingerprinting

Reproduction and extension of IEEE CNS 2019 work on fingerprinting encrypted voice-assistant traffic — end-to-end feature extraction and classification over network captures, with a public, reproducible repository.

ReproduciblePublic repository
PythonNetwork SecurityMachine LearningReproducibility

Currently

On the workbench

What I'm actively working on right now — no roadmaps, just the things in front of me.

~/muhammad-usama/now
$

air-gapped-delivery

Hardening zero-downtime rolling updates and offline image registries for Kubernetes in air-gapped environments.

active
$

on-prem-copilot

Extending the on-premise RAG co-pilot — sharper retrieval grounding and tool-calling for plant engineers.

active
$

telemetry-ingestion

Event-driven MQTT/Kafka ingestion for field sensor telemetry, tuned for reliability over throughput.

active
$

writing-series

A weekly technical writing series on industrial software engineering, starting with air-gapped Docker.

planning
git status