AI Research Scientist · Technical AI Governance · Trustworthy & Responsible AI
I'm Dr. Syeda Shafia Rubbani, an AI PhD (Heriot-Watt University, 2025) and FHEA (UK). My doctoral work applied Shannon's information theory to edge-AI protocol optimization; my current work focuses on the technical governance of multi-agent LLM systems: coordination risk, evaluation, privacy-preserving ML, and responsible deployment aligned with the EU AI Act.
How autonomous LLM agents interact, fail, and should be evaluated before deployment- grounded in building multi-agent pipelines end to end.
Differential privacy (DP-SGD, Rényi accounting), federated learning (FedProx, secure aggregation), and the trade-offs they impose on utility and latency.
Designing evals that measure what matters: hallucination scoring, high-stakes accuracy subsets, and LLM-as-judge pipelines with failure taxonomies.
Shannon capacity as a lens for protocol efficiency in constrained systems - a doctoral foundation I now bring to reasoning about AI system limits.
I treat deployable systems as part of my research identity: claims about AI governance should be grounded in what real systems actually do.
A five-agent cognitive decision-support platform (Interpret → Retrieve → Ideate → Evaluate → Recommend) that turns complex organizational problems into structured, evidence-grounded recommendations in ~40 seconds, with RAG citations and multi-tenant vector namespaces.
Python · FastAPI · ChromaDB · multi-agent RAGPrivacy-preserving federated platform for phishing & BEC detection: TinyLlama + LoRA multi-task heads trained with Opacus DP-SGD (ε < 5), federated via Flower FedProx with SecAgg+, plus an adversarial defense layer and threat-intel retrieval.
PyTorch · Opacus · Flower · FAISS · KubernetesAgentic clinical framework generating evidence-grounded medical reports from sparse inputs: a 9-agent LangGraph pipeline with ICD-10/SNOMED CT mapping, per-claim hallucination detection, PubMed/NICE citations, and human-in-the-loop physician review.
LangGraph · FHIR · SHAP · NLI scoringInteractive clinical reference dashboard for the ICBHI 2017 Respiratory Sound Database, built for the Edge-AI Smart COPD Mask project - FHIR-compliant patient summaries with explainability via SHAP and Grad-CAM.
FHIR R4 · SHAP · Grad-CAM · RailwayA BCI–LLM thought-communication prototype: an EEG signal pipeline (bandpass + ICA, FFT/P300 features, EEGNet intent classification) translated to natural language via an LLM — designed for patients with severe impairments.
EEG · EEGNet CNN · LLM translation · Node/ExpressA benchmark testing whether LLM systems give voters accurate election-logistics answers, with an LLM-judge scorer, a suppressive-vs-recoverable error taxonomy, and a high-stakes accuracy subset - a concrete study in governance-relevant evaluation design.
Python · LLM-as-judge · eval designDefined a protocol efficiency metric η = R / C normalized against the Shannon–Hartley bound, and validated it across CoAP, MQTT, AMQP and HTTPS through neuroevolutionary optimization and MIMO frameworks - raising measured capacity utilization from a 50–70% baseline to 70–90% with neuroevolutionary AI on resource-constrained edge devices.
A five-paper series is in preparation from the thesis, all single-authored. Targets below reflect current submission plans; statuses are stated honestly and updated as they change.
The η = R/C framework and cross-protocol empirical study for M2M edge communication.
Evolutionary search over protocol configurations against capacity, power, memory and handover objectives.
Theoretical treatment of capacity bounds under the mobility and MIMO regimes studied empirically in the thesis.
MIMO-based analysis of throughput, handover time, and stability during mobility.
Extends the thesis line toward learning-theoretic guarantees for remote estimation over constrained channels.
Includes an IoT testbed for smart-city sensing and cluster-based routing in mobile ad-hoc networks, among 6+ peer-reviewed articles from earlier academic roles.
Self-employed · Remote
Advising organizations on applied and generative AI; building and deploying the multi-agent, privacy-preserving, and health-applied systems above; continuing single-authored research toward publication.
School of MACS, Heriot-Watt University · Dubai
Doctoral research on information-theoretic protocol optimization; real-world edge AI & IoT experimentation (Contiki OS, Azure, GCP); led AED 50K government-funded project (Dubai Expo 2020); elected Postgraduate Research Representative for 100+ PhD scholars; winner, 3-Minute Thesis Competition.
University of Warwick · UK
Commercialization feasibility for AI-driven respiratory healthcare: 55+ structured stakeholder interviews across academia, industry and government; market feasibility report identifying a £300M opportunity, tied to a significant investment decision.
Heriot-Watt University · Dubai
Designed and delivered courses aligned with UKPSF standards; supervised 30+ final-year projects, six published in peer-reviewed venues.
MS Engineering College & Islamiah Institute of Technology · Bangalore
Taught 650+ students across core computer science; supervised 35+ student projects; co-authored conference papers with students and faculty.