TAHA TORBATI

A physics-informed surrogate that predicts a full set of aerodynamic coefficients directly from airfoil geometry, replacing expensive CFD sweeps across a wide flight envelope. Analytic derivative layers let the network respect the underlying governing equations while remaining fast at inference.

• Developed PINNs with CST parameterization trained on the NASA PALMO dataset, covering a broad operational envelope (Mach 0.25–0.9, AoA −20° to +20°).
• Achieved R² > 0.99 across lift, drag, and moment coefficients, with robust generalization to unseen geometries enabled by fast derivative layers.

A structural optimization framework that couples quantum sampling with classical metaheuristics and finite element analysis to search large truss design spaces more efficiently than either approach alone.

• Built the optimization pipeline combining quantum algorithms, classical metaheuristics, and FEA for sizing and topology.
• Demonstrated significant gains in computational performance and scalability on complex structural engineering problems.

A high-efficiency algebraic multigrid solver for the large sparse linear systems that arise when discretizing three-phase computational fluid dynamics, targeting both accuracy and throughput.

• Developed and implemented an algebraic multigrid (AMG) linear solver tailored for large-scale CFD systems.
• Enhanced simulation precision and computational throughput through highly optimized matrix operations.

A shaker-table test rig whose platform motion, cam geometry, and contact forces are derived from the governing equations — plus the signal-processing tools built alongside it. Each result grounded in the underlying mathematics rather than off-the-shelf tooling.

• 3D Motion Design: engineered the kinematic motion profiles, cam envelopes, and jack/cam forces for a shaker test table with SymPy, SciPy, and MSC Adams (June 2023).
• Signal Processing: personal projects from my Engineering Mathematics year — a Kalman filter, displacement from accelerometer data by drift-corrected double integration, and a Fourier audio compressor (July 2021).
• Machine Learning: my first ML project, written in high school at 15 — early diagnosis of Alzheimer's disease from EEG/ERP markers, reaching 85–91% accuracy.

Instructional design for physics and mathematics — building teaching methods and interactive, browser-based labs that let students learn STEM by running real experiments rather than memorizing.

• Designed physics and mathematics lesson plans built around how students actually learn, not just what they need to memorize.
• Led the ideation behind new, interactive, AI-driven ways to engage with lessons — making complex STEM topics feel intuitive instead of intimidating.
• Built interactive HTML-based virtual labs where students run real experiments and learn by doing — now used by 5,000+ students.
• Tailored teaching tools, formats, and instructional approaches to each lesson's specific needs, rejecting one-size-fits-all STEM education.

Owned the digital and product work behind Mana Battery's solar-and-storage offering — turning customer requirements into practical web software — and built AI systems that automate content creation.

• Designed and developed a web-based tool for solar panel system sizing: from a user's location and energy demand it proposes panel count and array area; conversely, from an available area it forecasts generation and runs the techno-economic analysis (payback and ROI).
• Designed and developed a web application that recommends and configures suitable UPS battery packages based on each customer's application and requirements.
• Designed and developed an AI-powered content-creation bot, automating image and text generation.
• Contributed to digital-transformation initiatives, identifying operational needs and converting them into practical software solutions.

Led Mana Battery's visual identity and creative direction across the brand's marketing and product communication.

• Directed the brand's visual identity and art direction across marketing and product communication.
• Produced design assets and campaigns, working primarily in Adobe Illustrator and the wider creative suite.

Applied machine learning and automation across industrial engineering and business workflows — from vibration-based predictive maintenance to risk modeling and pipeline automation.

• Developed predictive ML models correlating fatigue and vibration data to enable proactive maintenance strategies.
• Optimized sales and credit-risk workflows with ML-based customer segmentation and interactive dashboards.
• Automated critical engineering workflows with Python and SQL, reducing R&D testing duration by up to 70%.

Designed and taught advanced physics for high-ability students through camps, challenge exams, and interactive lessons built to reward deep understanding over memory.

• Ran hands-on camps (the Rope Workshop for Newton's laws and vectors) with students and their families.
• Created "work-exam" challenges: thermal insulation, energy machines, and a circuit escape room.
• Wrote notes and interactive activities, including a browser-based estimation and kinematics toolkit.

Led the problem-solving and doubt-clearing sessions for Design of Machine Elements I, curated the clearest explainer videos for each topic, and built interactive simulations for stress analysis.

• Worked full problem sets on the board and held one-to-one doubt-clearing hours.
• Curated a targeted video for each concept students found hard.
• Built interactive Mohr's circle, stress-concentration, and stress–strain simulations.