Software & AI Engineer • OCR/CV • Retrieval (RAG) • Backend
I build production-focused AI features that ship: document automation, retrieval pipelines, and services, measured, validated, and designed for reliability.
Based in Cape Town • Open to hybrid/remote roles
Computer vision, OCR, retrieval (RAG), and production-focused evaluation.
APIs, validation, data modeling, and maintainable services.
I’m comfortable moving across domains, from product-facing web systems to low-level reliability and production AI. This range helps me ship end-to-end solutions and communicate across teams.
Computer vision, document automation, RAG pipelines, evaluation, and safe deployment patterns.
APIs, data modeling, and systems that prioritize correctness, observability, and maintainability.
Latency-aware pipelines and hardware integration with an emphasis on signal integrity and robustness.
Algorithms, networking, and performance tradeoffs, useful when reliability and efficiency matter.
Filter by category, then open any card for a short case-study view.
Production OCR + LLM pipeline for document-to-structured-data automation.
RAG assistant connecting and matching African entrepreneurs via fast retrieval + LLM responses.
Sensor fusion + edge-AI foundations for real-time hazard detection.
Ongoing work on an AI-driven product with details limited by NDA.
Neural network image classifier with an end-to-end training and evaluation pipeline.
Full-stack web app emphasizing usability, data integrity, and maintainable APIs.
Hands-on systems work: scheduling simulations, protocol design, and integrity checks.
A small selection of public work you can review immediately. For private/NDA work, I share high-level details and can walk through specifics live.
A quick view of how my work evolved across domains. Details are intentionally high-level where needed.
Verified cloud learning and hands-on labs focused on practical, production-relevant skills.
Selected programs that strengthened leadership, communication, and technical execution.
Short, public-safe notes on how I build and ship reliable systems.
In production OCR, errors often look “plausible” and silently poison downstream systems. My default is validation-first extraction: explicit constraints, cross-field rules, confidence thresholds, and human-review hooks. The goal is not just extracting text, but producing outputs you can trust and audit.
A good RAG system is as much about “when to refuse” as it is about retrieval. I focus on deterministic indexing, conservative thresholds when evidence is weak, and clear provenance (links back to sources). This keeps the assistant helpful while staying honest and grounded.
What helps me move fast safely: reproducible environments, dependency hygiene, automated checks, and clear “definition of done”. Even lightweight CI, good error handling, and predictable releases reduce firefighting and increase delivery cadence.
I build end-to-end AI features that hold up in production: ingestion, validation, retrieval, model integration, observability, and delivery. My focus is correctness and reliability: clear interfaces, defensive checks, and measurable outcomes.
Software engineering, AI/ML engineering, and backend/cloud roles. Open to hybrid or remote opportunities.
University of Cape Town (Computer Science & Computer Engineering)
Note: this form opens your email client with a pre-filled message (no server, no data storage).