[Yaseen Khalil]|Computational Modeler & ML Systems Architect
> Exploring the mathematical architecture of intelligent systems. Bridging high-dimensional feature engineering with production data pipelines and autonomous AI integrations.
$ cat ./technical_matrix
Technical Matrix
skills.config
Computational Modeling
- >High-Dimensional System Dynamics
- >Semi-Tensor Product (STP) Algebra
- >Graph Neural Networks (GNNs)
- >Probabilistic Boolean Networks (PBNs)
- >Stochastic Simulations
- >Discrete Network Robustness
Data Science & Machine Learning
- >Deep Learning (Bi-LSTM, Attention Mechanisms)
- >High-Dimensional Data Analysis (KSG-KNN Estimators)
- >Clustering Algorithms (DBSCAN)
- >Neural Gene Expression Dynamics
Software Engineering & Architecture
- >Microservices & API Design (FastAPI)
- >Cross-Platform Mono-repos (TypeScript)
- >Relational Schema & RLS Security (Postgres)
- >Modular Frontend Component Architectures
- >High-Throughput Streaming Telematics
Languages (Primary)
- >Python
- >Go
- >Java
- >R
- >SQL
Languages (Secondary)
- >TypeScript
- >Mojo
- >KDB/q
- >CSS
Athletics & Lifestyle
- >Brazilian Jiu-Jitsu
- >MMA
- >Boxing
- >Weightlifting
- >Running
READY | 29 modules loaded
$ ls ./projects --verbose
Systems Architecture
- >Architecting analytical software to evaluate neural gene expression, focusing on high-dimensional system dynamics in developmental models.
- >Implementing KSG k-Nearest Neighbors (KSG-KNN) estimators to process complex datasets, extracting novel computational metrics to accelerate laboratory investigation.
Machine LearningNeural Gene ExpressionKSG-KNNSystems Biology
Research Ongoing
- >Architected a PyTorch-based Semi-Tensor Product Graph Neural Network (STP-GNN), engineering an algebraic transition operator using Khatri-Rao assembly to map discrete Boolean logic into a continuous, differentiable landscape scaled for 2²⁰+ state-space circuits.
- >Condensed a 30-variable module from an unstable 1,140+ attractor space down to 3 functional basins, executing 10,000-trajectory stochastic simulations to mathematically prove absolute structural robustness within the system dynamics.
- >Engineered gradient-directed perturbation algorithms to bypass invariant manifold bottlenecks, validating the architecture against discrete probabilistic baselines with <0.1% variance.
PyTorchSTP-GNNState-SpaceStochastic Simulation
ArXiv Publication Coming Soonview_research_code
- >Engineered and launched a cross-platform ecosystem using a unified TypeScript mono-repo and 110+ modular components, ensuring feature parity across web and native mobile applications.
- >Designed and normalized a PostgreSQL schema to support distributed scheduling, entity locations, and analytics with scalable RLS-based security.
TypeScript Mono-repoReact NativePostgreSQLRLS Security
- >Designed and deployed an end-to-end vehicle health monitoring neural network leveraging a bi-directional LSTM with attention mechanisms.
- >Improved anomaly detection accuracy from 40-60% to 87-95% through iterative retraining on 30K+ telemetry points every 10 minutes.
- >Architected a three-tier anomaly detection pipeline utilizing LSTM for sequential insights and DBSCAN for geographic hotspot clustering.
- >Built scalable FastAPI microservices on Railway with Supabase integration, exposing telemetry insights via an internal dashboard UI.
BiLSTMAttention MechanismsDBSCANFastAPISupabase
- >Developed responsive web components and interactive front-end features utilizing JavaScript and React.js.
JavaScriptReact.jsFrontend
Academic Project
$ git log --blog --oneline
Blog
Engineering a Cell: From 17,000 Dimensions to a Single Matrix