Skip to main content
PyConZA 2026

hlawusibisi

hlawusibisi avatar

Bio

Hlawulani Sibisi is a computer scientist, machine learning practitioner, and entrepreneur based in South Africa. He holds an Honours degree in Computer Science, specializing in Applied Mathematics, from the University of Johannesburg, and is currently pursuing a Master's degree in Machine Learning at North-West University, Potchefstroom.

He works as a Senior Software Developer at the Council for Scientific and Industrial Research (CSIR), South Africa's leading scientific research institution, where he applies his technical expertise to research-driven software solutions.
Alongside his professional and academic work, Hlawulani Sibisi is the founder of two startups tackling infrastructure gaps in underserved African communities. Delivery Quick Quick is a food delivery platform designed to serve small towns overlooked by major players like Uber Eats. General Maps is a GPS data collection initiative aimed at mapping rural villages across South Africa — and eventually the rest of Africa — that remain invisible on platforms like Google Maps.

Driven by a passion for using technology to close infrastructure and access gaps on the continent, he combines a strong foundation in applied mathematics and machine learning with hands-on entrepreneurship to build practical solutions for real-world African problems

Accepted Talks:

Simplifying the Noise: The Ramer-Douglas-Peucker Algorithm for GPS and Geospatial Data Processing

Raw GPS data is inherently noisy: a single vehicle trip can generate thousands of coordinate points, many of them redundant or slightly off due to signal drift. Before this data can be turned into usable road geometry, map features, or lightweight visualizations, it needs to be simplified without losing its essential shape. The Ramer-Douglas-Peucker (RDP) algorithm is one of the most widely used solutions to this problem, reducing a dense series of points on a curve to a minimal set that still preserves the original path's form within a defined tolerance.
This talk introduces the RDP algorithm conceptually — how it recursively identifies and retains only the points that contribute meaningfully to a curve's shape — and demonstrates its practical implementation using the Python rdp package. Attendees will see how a few lines of code can take a raw, noisy GPS trace and reduce it to a clean, minimal-point representation suitable for storage, rendering, or further processing such as map-matching.
Drawing on real-world use in geospatial data collection — including road digitization from vehicle GPS traces in underserved and unmapped regions — this talk will cover why point simplification matters for performance (smaller datasets, faster rendering, lower storage costs), for downstream processing (cleaner input for routing and map-matching engines), and for practical software engineering trade-offs (choosing an epsilon value, iterative vs recursive implementation, and where RDP fits in a broader geospatial pipeline). The session is aimed at developers and data practitioners working with location data, mapping, or any domain involving curve or trajectory simplification.

Our sponsors

Thinkst
NITheCS
OfferZen
Afrolabs Black Python Devs CoderLevelUp ESP Isazi Consulting Naughty Bean JetBrains
SerpApi