I build end-to-end systems where AI meets hardware — from sensor networks and computer vision pipelines to autonomous robots and industrial automation. My goal: machines that think, adapt, and work alongside humans.
// about me
I'm an aspiring Automation & Robotics engineer focused on AI-driven physical systems. My ambition is to design solutions where software intelligence is inseparable from hardware — from factory-floor automation to autonomous robots navigating real-world environments.
I approach engineering as a systems thinker: I care about the full pipeline from raw sensor data to deployment, not just the model or the script in isolation. That means thinking about communication protocols, data flow, deployment constraints, and real-world failure modes.
I'm also interested in simulation environments — including game engines — as training grounds for autonomous agents before they touch real hardware. Virtual environments are where I can iterate 100x faster and validate system-level logic safely.
// technical skills
// projects
Each project is built end-to-end — hardware to deployment — with full README, architecture diagram, and documented engineering decisions. Quality over quantity: 5 deep projects > 40 tutorials.
Python simulation of an autonomous cleaning robot (Roomba-style) that navigates and cleans 3 rooms — living room, kitchen, and bedroom — on a time-based schedule. Implements a hardware abstraction layer, state machine, and sonar-based obstacle avoidance.
Problem: Demonstrate HW ↔ SW communication and autonomous decision-making without physical hardware.
Solution: Scheduler triggers cleaning at set hours → controller transitions states (IDLE → MOVING → AVOIDING) → sonar drives obstacle logic → rooms cleaned in sequence.
Automated sensor network monitoring temperature, humidity, soil moisture, and light. Triggers actuators (fans, pumps, grow lights) based on configurable thresholds. Built to demonstrate real industrial automation concepts on accessible hardware.
Problem: Manual monitoring causes crop loss and energy waste.
Solution: ESP32 nodes + MQTT bus + automated control logic + live Grafana dashboard with SMS alerts.
ROS 2 mobile robot with SLAM-based mapping and autonomous navigation, extended with a YOLOv8 vision pipeline for real-time object detection and classification. Fully simulated in Gazebo — no expensive hardware required.
Problem: Robot must navigate unknown spaces and identify objects without human input.
Solution: ROS 2 nav stack + SLAM + camera feed through YOLOv8 inference node.
LSTM-based anomaly detection model trained on time-series sensor data to predict equipment failure before it happens. Deployed as a REST API with real-time scoring, tracked with MLflow, exported via ONNX for edge deployment.
Problem: Equipment failures are costly and hard to predict from raw sensor streams.
Solution: Streaming ingestion → LSTM anomaly scorer → FastAPI → dashboard alert.
A complete industrial inspection system: a camera captures frames, an AI model detects defects in real time, results flow through a message broker to a REST API, get stored in a database, visualised on a live dashboard, and trigger alerts when thresholds are exceeded. Every layer is containerised and deployed via CI/CD. This project demonstrates full systems engineering thinking — not just "a model that works".
Problem: Manual visual QA is slow, subjective, and doesn't scale.
Solution: Automated end-to-end pipeline from camera to actionable alert, with full observability and audit log.
A living repository of core robotics and control algorithms implemented from scratch in Python, with visualisations. Not a tutorial copy — each implementation is built to understand the math and trade-offs, then tested on simulated scenarios.
Includes: PID controller, Kalman filter, A* path planning, RRT, inverse kinematics, sensor fusion, basic reinforcement learning (Q-learning, PPO).
Format: Jupyter notebooks + Python modules + animated visualisations.
Reinforcement learning agent trained in a Unity ML-Agents open-world environment. The agent learns navigation, obstacle avoidance, and goal-seeking behaviours — core skills for autonomous systems — without needing physical hardware. Includes automated QA test scripts as a bonus.
Why game sim? Game engines allow 1000x faster iteration than physical robots. Skills learned here (sparse rewards, curriculum learning) transfer directly to real-world robot training.
// career roadmap
Python, Git, Linux, Arduino, basic electronics. First personal projects launched.
Build 5 strong end-to-end projects. ROS 2, computer vision (YOLO), AI pipelines (FastAPI + Docker), simulation (Gazebo). Contribute to open source robotics repos. Build the algorithm fundamentals repo.
TensorFlow Developer Certificate, ROS 2 course completion, Docker/Kubernetes basics. Merge PRs into robotics open source projects to show real collaboration.
Junior Automation / Robotics Engineer or AI Systems Engineer. Targets: autonomous systems, Industry 4.0, drone companies, or QA/test automation as a bridge role while building domain depth.
Lead AI-robotics systems from concept to production. Design the next generation of autonomous machines.
// contact
I'm actively looking for internships, collaborations, and junior opportunities in automation, robotics, and AI engineering. If you have a project, opportunity, or just want to talk engineering — reach out.