Computer vision, AI tracking and dataset engineering

About ASW Open R&D Internship Challenge

Africa Space Works is opening a hands-on internship challenge for students and early-career engineers who want to work on demanding space software, embedded systems, FPGA and AI problems. The program is built around practical engineering: internal repositories, cloud workspaces, selected FPGA cards, remote or office participation, weekly delivery discipline, and monthly technical reviews.

The common challenge is to build a safe test bench for a real-time camera-based tracking system. The system will use camera input, computer vision, FPGA and embedded processing, measurable validation, and controlled actuator-style outputs to detect and follow small moving targets in a controlled environment.

All tracks are connected. Track A enables execution, Track B proves performance, Track C builds the real-time hardware pipeline, and Track D builds the computer vision and AI tracking layer.

Role Overview - Track D: Computer Vision, AI Tracking and Dataset Engineering

As an intern on Track D, you will work as part of the computer vision and AI team. Your mission is to build the perception layer of the real-time tracking system: collect or generate useful data, detect moving targets from camera input, track them over time, estimate their trajectory, and reduce false positives.

This track must collaborate tightly with Track C. The goal is not only to create good Python notebooks or models, but to make the tracking logic practical, measurable and portable enough for FPGA and embedded engineers to implement, simplify or accelerate.

You will also work with Track B to turn tracking quality into repeatable evidence: datasets, replay videos, metrics, benchmarks and regression tests.

Key Responsibilities

• Build camera-based detection and tracking pipelines for small moving targets in a controlled test bench.
• Create and maintain datasets, annotation rules, synthetic examples, replay videos and evaluation protocols.
• Develop computer vision or AI approaches for target detection, classification, tracking, trajectory estimation and false-positive reduction.
• Compare simple classical computer vision methods with lightweight AI models and choose what can realistically move toward embedded or FPGA implementation.
• Collaborate with Track C to simplify, quantize, rewrite or restructure tracking code so that it can work with FPGA-friendly constraints.
• Work with Track B to turn tracking performance into repeatable tests, regression checks and measurable evidence.
• Compare model and algorithm variants using data-driven arguments: latency, accuracy, robustness, compute cost, memory footprint and hardware feasibility.
• Maintain daily visibility of progress and deliver concrete tracking demos, metric reports and integration notes during weekly and monthly reviews.

Requirements

• Strong interest in computer vision, AI tracking, signal processing, robotics perception or applied machine learning.
• Good working knowledge of Python.
• Willingness to learn and use Rust, C and shell scripting to support deployment, optimization, reproducibility and hardware integration.
• Ability to write clean experimental code and then simplify it into practical algorithms that another team can implement on embedded or FPGA targets.
• Baseline understanding of image processing, tracking, machine learning, datasets and metric analysis.
• Ability to work interdependently: using tests from Track B, infrastructure from Track A, and hardware interfaces from Track C.
• Excellent technical writing skills to produce clear notes, comparisons and technical trade-off explanations.
• Drive to execute practical engineering tasks over purely theoretical research.

Selection, Prize, and Hiring

• Fast Selection: CV screening, technical task, and interview.
• Cash Prize: Top contributors generating reproducible, integrated, and high-quality reusable engineering output may be awarded a cash prize.
• Hiring Path: The best profiles demonstrating strong technical judgment and integration skills may be offered full-time or longer-term roles at ASW.