Rising final-year Data Science undergraduate at GIKI with hands-on expertise in machine learning, deep learning, data mining, statistical modeling, and full-stack system development. Delivered production-grade platforms …
Rising final-year Data Science undergraduate at GIKI with hands-on expertise in machine learning, deep learning, data mining, statistical modeling, and full-stack system development. Delivered production-grade platforms including an AI-driven algorithmic trading system, large-scale e-commerce analytics pipelines, kernel-level Linux synchronization software, containerized CI/CD pipelines on AWS, and generative AI mixed-reality applications. Proven capability across predictive modeling, real-time data systems, MLOps, and quantitative analysis.
Delivered 8+ end-to-end Python, C++, MATLAB, Java, and automation projects across 5+ international clients spanning data analysis, scientific computing, and applied machine learning domains.
Built data processing pipelines, automated workflows, web scrapers, and analytical tools tailored to client business requirements and performance constraints.
Engineered C++ and Python applications including numerical computing utilities, simulation scripts, MATLAB models, and back-office automation systems for recurring client engagements.
Provided post-deployment support, debugging, performance optimization, and iterative enhancements ensuring consistent client satisfaction and repeat business.
Delivered 8+ end-to-end Python, C++, MATLAB, Java, and automation projects across 5+ international clients spanning data analysis, scientific computing, and applied machine learning domains, Built data processing pipelines, automated workflows, web scrapers, and analytical tools tailored to client business requirements and performance constraints, Engineered C++ and Python applications including numerical computing utilities, simulation scripts, MATLAB models, and back-office automation systems for recurring client engagements, Provided post-deployment support, debugging, performance optimization, and iterative enhancements ensuring consistent client satisfaction and repeat business
Tracked and analyzed operational and financial data in Excel / Google Sheets to support daily business decisions across inventory, sales, and production workflows.
Monitored inventory, sales flow, and production timelines using structured Excel-based reporting for daily and weekly stakeholder reviews.
Partnered with marketing and finance teams to improve workflow efficiency and performance tracking through shared reporting templates and review cadences.
Drove data-driven process improvements in a production environment by surfacing recurring bottlenecks and quantifying their impact on throughput.
Tracked and analyzed operational and financial data in Excel / Google Sheets to support daily business decisions across inventory, sales, and production workflows, Monitored inventory, sales flow, and production timelines using structured Excel-based reporting for daily and weekly stakeholder reviews, Partnered with marketing and finance teams to improve workflow efficiency and performance tracking through shared reporting templates and review cadences, Drove data-driven process improvements in a production environment by surfacing recurring bottlenecks and quantifying their impact on throughput
Designed and implemented a complete CI/CD pipeline for a containerized multi-service web application (six Node.js microservices and aReact frontend) integrating GitHub Actions with AWS EC2 and Elastic Container Registry.• Built automated GitHub Actions workflows completing the full build-to-deploy cycle in approximately 5-10 minutes per CI run, handlingbuild, unit testing, ESLint code analysis, Docker image build and push, and deployment to Dev, Testing, and Staging EC2 environments• Configured AWS infrastructure: six private ECR repositories, a wichat-ec2-ecr-role IAM role with cross-account policies, and a github-actions-ecr IAM user with scoped least-privilege permissions• Containerized six microservices (API Gateway, Auth, User, LLM, Question, Webapp) using Docker Compose with MongoDBpersistence and Prometheus and Grafana monitoring stack• Automated SSH-based deployment with ECR authentication, rolling image updates, and Gmail SMTP notification of build anddeployment status to stakeholders
Built an immersive single-page scrollytelling site narrating an NBA championship run with 13 total visualizations: a persistent Three.js 3Dcanvas hosting 3 cinematic scenes plus 10 interactive data charts, all driven by scroll progress.• Engineered a persistent Three.js WebGL canvas hosting 3 cinematic 3D scenes — a procedurally textured PBR basketball, a GLBplayer model ("The Contender"), and a drag-rotatable HDR-lit Larry O'Brien-style trophy• Implemented GSAP and ScrollTrigger-driven camera, lighting, and material transitions across all 3 scenes, delivering magazine-gradeeditorial feel and smooth narrative pacing• Built 10 interactive data visualizations driven by structured JSON datasets and synchronized to scroll progress: a full-bleed playoffbracket, a half-court SVG shot chart, Finals margin-of-victory line chart, era radar (small multiples), decade trio line chart, concentricround arcs, title distribution chart, title treemap, per-row sparklines, and a drought timeline• Integrated WebGL post-processing pipeline using UnrealBloom for cinematic lighting and JavaScript ES modules with GLTF andHDR/PBR pipelines for code organization
Built an end-to-end data mining and machine learning system to predict late deliveries and extract business insights from a large-scaleBrazilian e-commerce dataset.• Engineered features from 100,000+ Brazilian e-commerce orders (Olist dataset, 2016-2018) merged across 8 CSV tables; built time-based attributes (day / month / quarter / week), ratio interactions (freight_ratio), and binned numerics for downstream modeling• Performed exploratory data analysis across customer state, weekday, and product dimensions; quantified regional delay patterns, right-skewed price distributions, and outlier-driven delivery times via boxplots and correlation analysis• Applied Apriori association rule mining on discretized features; top rule (low total price + few items -> late delivery) surfaced 68% late-delivery rate at support 12% and lift 1.8, highlighting small-order operational inefficiencies• Trained Random Forest (0.82 accuracy, 0.79 F1) and XGBoost (0.84 accuracy, 0.81 F1) with hyperparameter tuning viaRandomizedSearchCV; segmented customers via K-Means (3 clusters, silhouette 0.55-0.60) with PCA-based 2D visualization
Designed and built a full-stack high-frequency trading system bridging machine learning classification with real-time FX execution viaMetaTrader 5, with stabilized WebSocket data streams and FastAPI inference at sub-millisecond latency.• Trained a regularized Logistic Regression model (L2 / LBFGS, 1,000 max iterations, balanced class weighting) on 110,414 FX samplesacross major currency pairs (EUR/USD, GBP/USD, USD/JPY) with an 88,331 / 22,083 (80/20) split and 30+ engineered technicalindicators (SMA, EMA, volatility, lagged returns), achieving 78.58% accuracy, F1 0.81 (Up) / 0.74 (Down), and 0.80 weighted precision• Developed a Python-based FastAPI backend with a single-entry controller (run.py) orchestrating database initialization, MT5integration, asynchronous WebSocket streams with 30-second heartbeat ping/pong and tick deduplication, and model inference at 0.70 threshold• Designed an end-to-end data pipeline with FIFO dataset management, historical data scraping, live trade capture, walk-forwardbacktesting with drawdown and maximum-drawdown tracking, SHA-256 model versioning, and automated retraining triggered when liveerror rate exceeds 60%
The AI-first Recruitment Operating System that unifies and automates the entire hiring lifecycle for modern enterprises worldwide.
Dr Job FZ LLC. 2026 © All Rights Reserved
