Where there is a problem, there is a symbolic constraint for it.
I build AI systems that are intelligent and verifiable. My work pairs LLMs with symbolic reasoning (Answer Set Programming with Clingo, Prolog, PDDL planning) so that agents and robots follow hard constraints instead of hoping the model gets it right.
- 🎓 Master of AI, RMIT University (December 2025, Distinction). Thesis: multi-head BERT combined with PDDL symbolic planning for natural-language-driven robot control
- 🎓 Bachelor's in Mechanical Engineering (Mechatronics), Lovely Professional University (2022), the hardware foundation behind my robotics work
- 🤖 Robotics Software Engineer at VxLab, RMIT (Melbourne). I build and maintain robot software (SLAM, motion planning, navigation, simulation) for the lab's UR5, TIAGo and Pepper robots, and mentor student capstone teams
- 🚀 Shipped: Aion (pip-installable scheduling agent) and ApplyVista (multi-agent RAG job assistant)
- 🔬 Exploring: symbolic guardrails for LLM tool calls over MCP, rule-aware robot navigation with ASP, WiFi CSI perimeter security, Bangla OCR with constraint decoding, and Rust for systems work
- 📍 Melbourne, Australia. Open to full-time AI/ML engineering roles.
- 💬 Ask me about neuro-symbolic AI, LangGraph, Clingo/ASP, ROS, and LLM pipelines
pip install aion-agent
A calendar agent that sends natural language requests through an LLM, then checks the result against logical scheduling constraints in Clingo (ASP) before anything is written to Google Calendar. The LLM proposes, the solver verifies.
Python · Clingo · Ollama · Google Calendar API · ASP
Production-deployed RAG pipeline for job applications.
A LangGraph multi-agent system that ingests job postings, retrieves relevant CV and portfolio context from a ChromaDB vector store, and drafts tailored application materials. Runs locally through Ollama.
LangGraph · FastAPI · Next.js · ChromaDB · Ollama · TypeScript
Visual robot programming for non-technical users.
A Blockly drag-and-drop interface for programming the TIAGo robot, with an embedded AI assistant that explains blocks, suggests workflows and dispatches ROS commands. React frontend, FastAPI + ROSBridge backend, containerised with Docker.
React · Blockly · FastAPI · ROS · ROSBridge · Docker · Ollama
| Domain | Tools |
|---|---|
| Neuro-Symbolic AI | Clingo (ASP) · Prolog · PDDL · pyswip · LLM + solver integration |
| LLM Pipelines & Agents | LangGraph · LangChain · Ollama · RAG · ReAct · MCP |
| Robotics Software | ROS / ROS2 · ROSBridge · SLAM & mapping · Motion planning · Navigation · Kinematics & robot control · Simulation |
| Robot Platforms | TIAGo · UR5 · Pepper (NAOqi) |
| Robot Perception | Computer vision for robots · OpenCV |
| Embedded & Hardware | Arduino · sensors · microcontrollers |
| Full-Stack AI Apps | React · Next.js · TypeScript · FastAPI · ChromaDB · Docker |
| ML / Deep Learning | PyTorch · TensorFlow · Transformers · scikit-learn |
| Data & Vision | OpenCV · pandas · NumPy · Jupyter |


