Saeed Parvar Ph.D.

Author of Algorithmic Trading Strategies: A Step-by-Step Guide into the World of Quants, covering 70+ quantitative trading strategies and practical market analysis techniques. Developed quantitative FX models using high-frequency data (EUR/USD, EUR/GBP, AUD/USD) to analyse forward returns, volatility regimes, and market dynamics, supporting pricing and risk-aware decision making. Built and backtested classification- and regression-based models for financial market prediction; evaluated trading signals using rigorous backtesting and walk-forward analysis while mitigating overfitting through disciplined optimisation and parameter tuning. Designed delta-aware, risk-adjusted signal frameworks by analysing directional accuracy, confidence filtering, and signal strength across multiple horizons. Developed a modular algorithmic trading platform supporting crypto, FX, equities, and commodities, integrated with Interactive Brokers and Binance, incorporating 130+ strategies and improving execution and decision latency by ~25%. Implemented integrated backtesting, forward-testing, and adaptive strategy selection to maintain robustness across changing market regimes. Built time-series forecasting pipelines using ARIMA, Prophet, Exponential Smoothing, XGBoost, LSTM, and Gaussian Processes, improving forecast accuracy by ~12% and supporting pricing and allocation decisions. Conducted large-scale data analysis on S&P 500 equities, crypto assets, and Fama–French factors; engineered features on liquidity, returns, and factor betas, improving strategy decision accuracy by ~14%. Applied unsupervised learning (K-Means clustering) for monthly asset grouping and strategic allocation, increasing portfolio turnover efficiency by ~20%. Performed portfolio risk analysis and optimisation, focusing on robustness, diversification, and risk-adjusted performance.

- Core Competencies ML: Random Forest, XGBoost, PCA, CNNs, NLP. Predictive modelling across regression, classification, clustering, and time-series. Delivered churn prediction, demand forecasting, and optimisation with measurable ROI. - Business Insights & Optimization Horse-race outcome models (Random Forest, XGBoost, ANN): R² improved from ~20% to 80%+ via feature engineering, ensembling, and tuning. Churn prediction: +15% retention-strategy effectiveness (ensemble ML). Citibike NYC demand forecasting (ARIMA): 25% cost reduction via fleet optimisation. - Regression & Classification Salary, profit (R&D/marketing/admin spend), and social-ad purchase prediction; NLP sentiment analysis on reviews at 85% accuracy. - AI Document-Intelligence Pipeline (personal project) End-to-end pipeline to ingest, normalise, search, and reason over a large heterogeneous corpus. Microsoft Graph API ingestion; ETL across PDF/Word/text; lexical retrieval with context/token engineering to fit large corpora into limited context. Citation-grounded generation with strict anti-hallucination discipline (every claim traced to a named source), structured extraction, agentic multi-step reasoning, and isolated knowledge bases with clean provenance. Hands-on across retrieval, grounding, agentic reasoning, and data governance. - Clustering & Dimensionality Reduction Mall customer segmentation (behavioural clusters); wine classification with PCA. - Deep Learning & Advanced Techniques Churn prediction (ANN); CNN image classification; RL for ad-CTR optimisation (+20% ROI); market-basket analysis via association rules.

* Project Management Assistant — technical progress reporting, milestone tracking, and sponsor presentations for an industry-funded multiphase flow simulator. * Developing ML surrogate models to accelerate phase-equilibrium computation — training learned approximations of the flash solver to cut iterative cost, targeting real-time flow simulation. * Building a GPU-accelerated thermodynamic engine providing phase-behavior and fluid-property closures for multiphase flow / CFD simulation of reservoir fluids — CUDA C++ (NVCC / C++17), from first principles. * Implemented cubic equations of state (SRK / PR) with multiphase flash, stability analysis, and phase-envelope construction — the numerical core flow simulators depend on for accurate results. * Engineered performance-critical CUDA kernels (FMA intrinsics, memory-safe large allocations, device-side logging) — the same GPU/HPC foundation underpinning AI-accelerated scientific computing. * Built export pipelines feeding industry-standard flow simulators for downstream production and flow modeling, Pipesim, PVTsim, Ledaflow. * Validated against a commercial reference to production tolerances via finite-difference ground-truthing and thermodynamic consistency checks.

Developed analytics workflows to measure how website UI/UX changes impact KPIs such as visits, bounces, bounce rate, pageviews, and conversion-related metrics. Built GraphQL data pipelines to fetch, aggregate, and prepare website analytics data for causal inference and anomaly detection. Implemented Difference-in-Differences and event-study models to estimate the causal impact of UI changes using treated/control page groups. Developed time-series anomaly detection and point anomaly analysis to detect abnormal KPI behaviour and identify anomalous dates. Built root-cause analysis modules using chi-square tests, rate-based significance testing, and correlation-based driver analysis to explain KPI shifts across categories/dimensions. Integrated model outputs with ClickHouse and improved backend reliability through validation, error handling, testing, and GitHub-based code review.

Developed an AI-powered tutor and teacher-assistant agent for Python education, monitoring student progress, analysing performance trends, and generating personalised feedback reports. Designed a learning analytics framework covering Mastery, Resilience/Grit, and Conceptual Understanding, using behavioural signals such as time taken, retries, hint usage, stuckness, and task completion. Implemented conceptual gap analysis across six pedagogical pillars — ITEM, STRUCTURE, PURPOSE, REASON, RELATION, and APPROACH — to identify why students struggle with programming tasks. Built student and teacher analytics including mastery maps, resilience radar charts, class performance heatmaps, hard-topic rankings, and targeted intervention recommendations. Added adaptive feedback mechanisms using Socratic hints, hint-effectiveness tracking, adaptive difficulty levels, and personalised follow-on diagnostic questions.

* Securing EuroHPC Grant Access for advanced computational resources, 20,000,000 (Core/Hour). * Contributed to the development of advanced algorithms totaling over 10,000 lines of code in Python, MATLAB, and Fortran, facilitating the simulation of complex systems. * Skilled in data visualization, interpretation, and the creation of comprehensive technical reports, effectively communicating complex findings to both technical and non-technical audiences. * Implemented a range of machine learning techniques—including PCA, FFT, regression models, and neural networks—to analyze simulation data and extract actionable insights for system optimization. * Collaborated with leading academic institutions such as Strathclyde University (UK), Concordia University (Canada), Technical University of Madrid (Spain), and the University of Porto (Portugal) on over five high-impact projects, demonstrating exceptional teamwork and collaborative problem-solving. * Led and mentored a team of engineers, overseeing simulation tasks and algorithm development, fostering a high-performance, results-oriented team culture. * Presented key research findings and innovations at reputable conferences, contributing to the broader scientific community and showcasing expertise in computational fluid dynamics and machine learning. Key Projects: * Simulation of ElastoViscoPlastic Fluid Flows Past a Porous Medium (Collaboration with Strathclyde University, UK). Heat Transfer in ElastoViscoPlastic Fluids within a Cavity (Collaboration with Concordia University, Canada). * Machine Learning Applications in POD and HoDMD (Collaboration with Technical University of Madrid, Spain). * Simulation of Non-Newtonian Fluid Flow Past a Circular Cylinder (Collaboration with Luca Brandt Group, KTH). * ElastoViscoPlastic Fluid Flow Past a Confined Cylinder (Elastic Turbulence) (Collaboration with University of Porto, Portugal.

* Utilized advanced computational fluid dynamics packages such as Ansys Fluent and the RheoFoam toolbox of OpenFOAM to simulate and solve complex engineering problems. * Identified technical solutions and optimized simulation tools to meet multiple criteria, enhancing efficiency and accuracy. * Collaborated with team members to manage project timelines and allocate resources effectively, ensuring successful implementation of complex projects. * Conducted data visualizations and analyses to generate detailed technical reports, providing valuable insights for decision-making and project advancement. The numerical and analytical study of • Boundary (and heat transfer) • Mixing layers, • Jet flow of FENE-P fluid by RheoFoam toolbox of Openfoam (Collaboration with the Instituto Superior Técnico, Portugal).

* Developed a quantitative and statistical model (Large Eddy Simulation) for turbulent flow in collaboration with Instituto Superior Técnico, Portugal, utilizing advanced computational methods. * Designed and implemented algorithms totaling over 20,000 lines of code in Fortran, MATLAB, and C++ to simulate turbulent flow on various High-Performance Computing clusters. * Conducted comprehensive qualitative, quantitative, and statistical analyses on vast amounts of simulation data, extracting valuable insights for scientific advancements. * Identified and implemented advanced data visualization methods to enhance the interpretation of high-fidelity data. Selected Projects: * Developing a Large eddy simulation model for inhomogeneous wall-free turbulent viscoelastic fluid flows (Collaboration with the Instituto Superior Técnico, Portugal). • Renewable energy systems a. Wind Energy in Urban Environment: concepts, technology, and potential. b. High-temperature thermal energy storage for thermoelectric solar power plants c. Energy storage for thermoelectric solar power plants d. The production and trade of wood pellets e. Importance of solid oxide fuel cells f. The Energy Return of Investment of Anaerobic Digestion g. The Production of Bioethanol; Pros and Cons h. Tidal power system ** Conducted comprehensive studies on the concepts and technologies of various renewable energy sources to assess their potential for energy generation.

* Conducted simulation of turbulent non-Newtonian flow in a rod-roughened channel, contributing to the understanding of complex fluid dynamics phenomena. * Developed over 2,000 lines of code in C++ for post-processing of simulation data, enabling qualitative, quantitative, and statistical analyses to extract valuable insights. Pipesim, Olga, PVTsim