AI & Machine Learning Developer and Researcher · Peshawar, Pakistan
A computer scientist working at the intersection of applied machine learning, graph-based modelling, and quantitative research. I build production AI systems and publish peer-reviewed work that bridges rigorous theory with measurable real-world outcomes.
I currently serve as AI Lead at Vertexons, where I architect production LLM pipelines and MLOps infrastructure, and as Lead Researcher at Digging Deeper Insight UK Ltd, delivering AI-driven analytical work for a UK research organisation. In parallel, I develop a production-grade MetaTrader 5 Expert Advisor as an independent quantitative developer.
My research spans three converging frontiers: spatiotemporal Graph Neural Networks for climate-driven disease surveillance (1st Place, KMU 2026); deep reinforcement learning for closed-loop AI therapeutics, where a Soft Actor-Critic agent achieved zero severe hypoglycaemia versus a 4.5% failure rate in standard insulin protocols; and statistical evaluation frameworks for large language model factuality and reliability.
I am preparing graduate study in artificial intelligence and autonomous systems, with the long-term ambition of leading a research-driven technology company at the intersection of quantitative finance, AI safety, and graph-based reasoning.
Digging Deeper Insight UK Ltd
Conducting applied, policy-relevant research for a UK-based professional research organisation. Apply AI and data science methodologies to questions across policy, social, and economic domains; author structured deliverables (reports, evidence frameworks, analytical briefs) within professional project timelines.
Vertexons
Lead backend development and deployment of production ML systems across cloud environments. Design scalable FastAPI endpoints integrating LLM services for low-latency inference, and establish system standards across observability, MLOps, CI/CD, and human-in-the-loop feedback.
Self-Employed · MT5 Expert Advisor
Built a production-grade MetaTrader 5 Expert Advisor in MQL5 with statistical feature engineering, regime-aware signal integration, basket take-profit logic, trailing-stop mechanisms, and configurable risk controls. Validated robustness via Bayesian methods and the MT5 Strategy Tester across multiple market regimes.
AI-Powered Timetable System, IM|Sciences
Engineered a full-stack AI timetable platform serving 1,200+ students with intelligent routing between static schedule data and LLM-based responses. Built FastAPI REST services, a serverless backend proxy for API security, model failover, and streaming, plus an admin analytics dashboard and offline-capable PWA.
AI Thumbnail Generator
Co-founded an AI-powered thumbnail-generation platform selected for incubation at the National Incubation Center. Built the FastAPI backend for a generative AI service with third-party API integration, rate limiting, and serverless deployment.
Institute of Management Sciences (IM|Sciences)
Designed a multi-dimensional statistical framework for benchmarking LLM factuality, consistency, and reliability. Conducted model-benchmarking experiments; co-authored research methodology with manuscript in preparation for journal submission.
PTCL Group · withML
Optimised data-processing workflows for technical reporting systems across large-scale telecom infrastructure datasets at PTCL Group. Contributed to applied ML tasks at withML — data preprocessing, model experimentation, and evaluation workflows.
Bright Network · IEUK
Architected a 6-month product roadmap for the RunWize fitness platform, driving a 20% user-engagement increase. Synthesised behavioural insights from 3,900+ users to prioritise features and propose gamification initiatives.
Interpretable spatiotemporal Graph Neural Network with integrated causal discovery, modelling relationships between climate variables and regional disease outbreak patterns across Pakistan. Validated against epidemiological ground truth.
Reduced post-Iftar hyperglycaemia by 35 mg/dL; achieved zero severe hypoglycaemia versus 4.5% in standard protocols; outperformed clinical sliding-scale protocols in 92% of simulated scenarios.
Production-grade MT5 trading system with real-time market-data pipelines, statistical feature engineering, basket take-profit logic, regime-aware signal integration, and Dockerised CI/CD on AWS.
Full-stack platform with intelligent routing between static timetable data and LLM-based responses. Serverless backend proxy for API security, model failover, streaming; admin analytics dashboard; offline-capable PWA.
Statistical evaluation framework for benchmarking large-language-model factuality and reliability across multiple dimensions. Research Assistantship, IM|Sciences. Co-author.
Quantitative asset-allocation algorithm using Bayesian optimisation to maximise risk-adjusted returns, with acquisition functions (EI, UCB) for efficient hyperparameter search. Robust out-of-sample performance vs baseline.
Institute of Management Sciences (IM|Sciences)
GPA: 3.76 / 4.00 · Medium of instruction: English. Coursework: Machine Learning, Deep Learning, Algorithms & Data Structures, Software Engineering, Database Systems, Probability & Statistics.
AtomCamp
Intensive training in statistical modelling, ML deployment, computer vision, NLP, and LLM fine-tuning.