I build nested, multi-timescale learning systems — models that learn and synchronise at different frequencies across layers, agents, and devices.
- 🎓 MSc Artificial Intelligence, Brunel University London
- 💼 AI/ML Software Engineer, Electronic Media Services — London
- 🔬 AI Researcher & Co-Investigator, Telecommunications Research Lab (TRL), IBA Karachi
- 🚗 AI/ML Engineer, Curium — multi-sensor calibration for autonomous perception
- 🛡️ Formerly AI Research Engineer, National CERT (Govt. of Pakistan)
- 🌍 Publishing with collaborators across the UK, US, UAE, Pakistan, Saudi Arabia & Oman
- 💬 Ask me about agentic AI, digital twins, federated learning, multi-timescale RL, LLM orchestration
AI/ML Software Engineer · Agentic AI & Digital Twin Technologies
|
Designing autonomous multi-agent systems — planning, tool use, and orchestration layers that let LLM agents reason, delegate, and act reliably in production environments.
|
Building live virtual replicas of physical systems — sensor-driven models that simulate, predict, and optimise real-world behaviour in real time.
|
 Nested Learning Layer-wise multi-frequency synchronisation across timescales |
 Federated & Privacy Federated unlearning, differential privacy, data-rights compliance |
 Agentic & Efficient LLMs Orchestration, semantic caching, complexity-aware routing |
 6G & Network AI Multi-agent RL for slicing and Open-RAN resource allocation |
 Clinical & Biomedical AI ECG classification, vessel segmentation, cross-hospital prediction |
 Security & Forensics Ransomware detection, intrusion detection, multi-agent forensics |
Full list on Google Scholar · ResearchGate
- Nested Multi-Agent Reinforcement Learning for Adaptive Resource Management in 6G Network Slicing — IEEE Open Journal of the Communications Society, 2026
- HADES — Hierarchical Autonomous Digital Evidence System via Skill-Driven Multi-Agent Orchestration · WI 2026, Linz
- Hierarchical Bayesian Nested Optimisation for Uncertainty-Aware Resource Allocation in 6G Open-RAN
- Training-Free Coarse Point Cloud Registration via Axis-Projected Surface Area Matching · ICCVDM 2026
- TEMPEST — Temporal Execution Modeling for Pre-Encryption Ransomware Detection
- Nested Federated Learning — Layer-Wise Multi-Frequency Synchronization · Cognitive Computation
- Nested Learning for Adaptive-Fidelity ECG Classification — Matryoshka representations across the wearable-to-cloud continuum · IEEE JBHI
- Nested Learning for Multi-Timescale EV Charging Coordination — physics-informed deep RL with hard voltage guarantees · IEEE Trans. Smart Grid
- DP-f-FUM — Differentially Private Federated Unlearning · FLTA 2026
- NestAnt — Nested-Learning Surrogate for Continual Multi-Band Antenna Design · IEEE GLOBECOM 2026
| Project | What it does | Stack |
|---|---|---|
| LLM-Powered SIEM NCERT, 2025 |
National SOC pipeline — Kafka ingestion → quantized inference → automated alert routing, 25M+ logs/day across 200+ federal orgs | PyTorch HuggingFace DeepSpeed Kafka K8s |
| Enterprise Multi-Agent RAG | Hierarchical agent orchestration + hybrid retrieval; accuracy 76% → 89% at sub-200 ms latency | gRPC Llama.cpp Qdrant Elasticsearch Redis |
| Thermal Industrial Monitoring | ViT+CNN with spatio-temporal GNNs; 90–95% accuracy via contrastive learning | PyTorch timm PyTorch Geometric |
| ZazuAI — Meeting Intelligence | Dual-pipeline speech: real-time (<500 ms) + high-accuracy post-processing (96%) | Whisper.cpp DIART WhisperX |
| 🎯 | Selected Participant, CISPA Summer School 2026 — Federated Learning track (Helmholtz Center for Information Security) |
| 🇵🇰 | Selected to Pakistan's National AI Research Team, NCERT (2025) |
| 📜 | Letter of Appreciation, Director General NCERT — national SOC initiative (2025) |
| 💰 | NESCOM Research Funding — intelligent UAV communications (2024) |
| 🎤 | Industry Trainer, IBA CICT — Agentic AI & Intelligent Automation program |
Open to research collaborations in nested learning, federated systems, agentic AI, and digital twins.
