I'm an AI Engineering undergraduate at Mohamed bin Zayed University of Artificial Intelligence, focused on LLM systems, agentic workflows, and model evaluation. I build data pipelines, PyTorch prototypes, and reliability-oriented AI applications, with interests in statistical learning, sequence modeling, decision systems, NLP, and reliable LLM deployment.
- LLM systems, agentic workflows, and model evaluation
- controlled generation and inference-time policy design
- reliable evaluation under latency, token, and compute constraints
- hidden-state geometry and mechanistic interpretability
- sequence modeling and NLP for classical Chinese text
| Project | Status | What it shows |
|---|---|---|
| Deadline-Valid Replay: Anytime-Valid Testing for LLM Inference Policies | First author, ICML 2026 Hypothesis Testing Workshop | Statistical evaluation methods for LLM inference policies under latency deadlines, separating leaderboard gaps from certified policy-level advantages. |
| Metric Choice Determines Semantic Geometry in LLM Hidden States | First author, ICML 2026 Mechanistic Interpretability Workshop | Layerwise hidden-state geometry analysis showing raw L2 expansion is driven by norm growth while centered cosine better reveals semantic, topic, and category structure. |
| Chronos: Natural-Language Scheduling | Natural-language scheduling product | Converts natural-language scheduling requests into Google Calendar and Outlook events, including preview, edit, delete, multi-event requests, recurring schedules, relative dates, timezone handling, undo/redo, command persistence, auth, and API validation. |
| RLM: Defect-Guided Verifier Search for Controlled LLM Generation | Ongoing for ICML 2027 | Controlled-generation system combining RLM-style hidden-context access with Defect-Guided Verifier Search over visible candidates, with typed schemas for facts, forbidden claims, constraints, verifier gates, repair targets, candidate logs, branch budgets, and token accounting. Repository link withheld while the project is being prepared for an ICML 2027 submission. |
| AI Archaeology of Chinese Songci | Ongoing for ISMIR 2027 | Songci NLP pipeline over classical poetry JSON corpora, extracting cipai, line length, rhyme position, ping/ze tone, and rhythmic features into PyTorch sequence-modeling inputs. Repository link withheld while the project is being prepared for an ISMIR 2027 submission. |
| SymStep: Symbolic Step Verification for Logical Reasoning | Coauthor, IJCAI/ECAI 2026 NILA Workshop | Step-level symbolic verification for LLM reasoning, using constraint propagation to check atomic deductions and strengthen constraint-dense logical reasoning. |
- Languages: Python, SQL, C++
- ML/AI: Machine Learning, Deep Learning, LLMs, Agentic Systems, LLMOps, Model Evaluation, PyTorch, NumPy, Pandas
- CS: Data Structures and Algorithms, Operating Systems, Software Engineering, JavaScript, React
- Tools: Git, GitHub, VS Code, Jupyter, LLM APIs
- B.S. in AI Engineering, MBZUAI, expected Fall 2029
- MBZUAI Dean's List
- Sheikh Tahnoon bin Zayed Scholarship in AI Excellence and Pioneer Scholarship
- President, MBZUAI Climbing Club
- Founding student peer tutor for Calculus and Linear Algebra