Blessing Hlongwane
Data Scientist — looking for Software Engineering & Full-Stack roles
📍 Cape Town / Johannesburg · University of Cape Town
Data Scientist with AWS cloud experiencestudying Computer Science & Applied Statistics at the University of Cape Town. Looking for a full-stack software engineering or development role in Cape Town or Johannesburg.
My projects are the best way to see what I can do — I've built full-stack applications across data pipelines, web platforms, and systems engineering.
- Improved categorization algorithms and monitored AWS EC2 instances, CloudWatch dashboards, shell scripts, and Lambda triggers to keep data pipelines running efficiently
- Conducted data analytics uncovering insights into customer spending behaviour and market share trends
- Built operational dashboards in Streamlit and Metabase; maintained and iterated on a production forecasting model pipeline
- Automated infrastructure tasks using shell scripts and Lambda functions; tracked anomalies via CloudWatch alarms and S3-backed data flows
- Evaluated student assignments and provided constructive, detailed written feedback
- Supported students in understanding core CS concepts including algorithms and data structures
- Designed and delivered lessons on HTML, CSS, and JavaScript fundamentals to high school students
- Facilitated hands-on coding exercises to build foundational programming intuition
Full-stack event calendar application with RSVP functionality, automated guest email invitations, AI-powered facts and quotes per RSVP, and confirmation email delivery.
- Event creation with full details, automated guest email invitations via Gmail SMTP with unique RSVP links per guest
- OpenAI integration serving personalized AI-generated facts and quotes on each RSVP confirmation page
- SQLAlchemy ORM on SQLite with Pydantic data validation; deployed on Render
Android peer-to-peer messenger that routes messages over Bluetooth mesh connections — fully offline, no internet required. Final year group project at UCT, scored 80%+.
- Bluetooth nodes relay messages through the mesh — works in areas with zero connectivity
- Android app with chat UI; messages routed through intermediate devices acting as nodes
- Collaborative group project with version-controlled development workflow
A concurrent Java simulation of a medley swimming relay race. Multiple threads represent swimmers racing in parallel through a shared grid — a coursework project exploring concurrency primitives, thread synchronisation, and shared-state management.
- Threads represent individual swimmers; shared stadium grid managed with synchronized access
- FinishCounter uses concurrent primitives to track and display race results in real time
- StadiumView renders a live ASCII/graphical view of the race as threads update position
Optimized a sequential Abelian sandpile model image generator by introducing parallelization using Java's Fork/Join Framework. Achieved a 3× average runtime improvement on large datasets.
- Fork/Join recursive decomposition of the sandpile stabilisation computation
- 3× average speedup on large datasets via work-stealing thread pool
- Visual output comparing sequential vs parallel execution across dataset sizes
A Python desktop image viewer with a custom GUI — supports directory browsing, image editing, a favourites system, and keyboard navigation.
- Directory picker for browsing local image folders with thumbnail preview
- Image editing utilities: crop, resize, rotate, and filter operations
- Persistent favourites list stored locally for quick access to saved images
A Python platformer game inspired by Super Mario — built to explore game physics, sprite animation, and tile-based level design.
- Platform physics: gravity, collision detection, variable-height jumping
- Sprite animation state machine (idle, run, jump) and tile-based level rendering
- Featured on LinkedIn — received strong community engagement
AI-powered email automation tool using n8n workflow engine and OpenAI GPT. Automatically fetches, processes, and summarizes daily emails into digestible highlights.
- n8n workflow triggered on schedule to fetch inbox via Gmail API
- OpenAI GPT condenses long email threads to key action points
- Delivered summaries to configured notification channel automatically
Live crypto trading bot for Polymarket 5-minute UP/DOWN binary markets. Integrates Coinbase WebSocket spot data, Deribit implied volatility, Polymarket CLOB order books, and Chainlink/Polygon resolution data into a single low-latency execution engine.
- Strategies: probability signals, order-flow imbalance, microprice, dynamic position sizing, arbitrage, market-making
- Execution layer: CLOB order signing, FOK/FAK/GTC order handling, pre-signed orders, latency-aware buy-path optimisation
- Risk tooling: exposure limits, budget caps, automated merge/redeem workflows onchain, balance recovery, post-resolution settlement
Data collection and analysis pipeline for live Polymarket and Coinbase order-book data. Used to validate strategies before deploying to the live trading bot — simulation-first approach.
- Tick data collection from Polymarket & Coinbase CLOB order books via WebSocket
- Probes for lead-lag behaviour, fill quality, signal accuracy, flip detection, and latency-sensitive execution decisions
- Results drove trading rules, order types, sizing logic, and risk controls in the trading bot