B.Sc. Computer Science Student · RWTH Aachen University

Salifou Affoh

Building AI products end to end, grounded in rigorous software engineering.

Focus

Software Engineering

Building and shipping production software as part of a team: implementing features end-to-end in C++ and React, working from protocol specs to defensive, well-tested code, with the Git and CI/CD workflows that real collaborative engineering runs on.

C++React / React NativeGitDockerGitLab CI/CD

AI Engineering

End-to-end LLM applications, not just model calls: retrieval-augmented generation with citation verification, tool-calling agents, rule-based recommendation engines, and automated evaluation and red-teaming.

RAGLangChain / LangGraphAgentsEvaluation & Red-TeamingNeo4j

Experience

March 2025 – September 2026

Working Student - Software Engineer

Qubicast

Developed modern user interfaces with React Native and TypeScript, and designed interactive network visualizations, for QubiSet — a cross-platform tool for configuring, controlling, and monitoring networked professional lighting equipment.

Implemented lighting-control protocol specifications (Art-Net, sACN, DMX/RDM, LLRP) from scratch in C++ and integrated them into the existing server-side protocol stack.

Built state-management, API-integration, and client-server communication components, plus features for managing devices, groups, and networks.

Wrote and maintained automated unit and integration tests, and stabilized existing functionality and CI pipelines through systematic debugging.

Applied extensive defensive programming practices to handle real-world device behavior, since manufacturer hardware frequently deviates from spec or documented behavior.

Contributed to licensing, demo-mode, and usage-analytics features, working in a full-stack, cross-functional team in a startup environment.

C++ReactReact NativeTypeScriptElectronGit

Skills

Languages

C++PythonTypeScript

AI & ML Engineering

OpenAI APILangChainLangGraphLangSmithNeo4jFastAPIOpenCVPyTorchPandasNumPy

Web & DB

ReactNext.jsNode.jsExpressPostgreSQLmongoDBredisNeo4j

Tools

IntelliJVimGitGitHubDockerLinuxJupyter Notebook

Spoken Languages

FrenchNative
GermanC1
EnglishC1

Soft Skills

Problem-solvingCritical ThinkingTeamworkTime ManagementContinuous LearningResult OrientedOwnership & Responsibility

Projects

StudyRAG — Citation-Backed Q&A & Elective Recommender

Visit site

Built a citation-backed Q&A system over a university's degree regulations: hybrid retrieval with LLM reranking, a tool-calling agent for live lookups the indexed corpus can't cover, and automated citation and version-conflict checking so answers are never just a confident guess. Layered a career-goal-driven elective recommender on top of the same pipeline, backed by a Neo4j prerequisite graph and real job-market evidence, evaluated end to end with a 27-persona test suite and a scheduled prompt-injection red-team suite, with LangSmith tracing for production visibility.

9 career tracksCitation-verified RAGTool-calling agent27-persona eval suiteRed-team tested
PythonFastAPIOpenAI APILangChainLangGraphLangSmithNeo4jNext.jsRedisDocker

Real-Time Fly Embryo Detection and Analysis Dashboard

Built a real-time OpenCV pipeline (preprocessing, contour detection, object tracking) to automate embryo detection and tracking in live video data. Designed a real-time controls dashboard for adjusting algorithm parameters with live feedback, optimizing detection accuracy. Additionally, collected and visualized evolution metrics to monitor and analyze embryo activity.

Real-time OpenCV pipelineLive parameter-tuning dashboardEvolution-metrics visualization
PythonOpenCVPandasPanelHolovizGit

Graph Algorithm Visualization

Visit site

Contributed to the backend development of a graph algorithms visualization web app designed to assist RWTH professors in teaching graph algorithms. Focused on implementing exercise generation and grading features, collaborating with cross-functional teams in an agile environment to plan features, conduct code reviews, and ensure seamless codebase integration.

Exercise generation & gradingUsed by RWTH course staffAgile, cross-functional team
ReactJSTypeScriptWebAssemblyRustGitGitlab CI/CD

FrontendFight: Multiplayer CSS battle

A multiplayer CSS learning game designed to make coding both educational and engaging. Built with Angular and Node.js, the game allows players to join lobbies and compete in solving CSS challenges. A pixel-level image similarity library ensures accurate evaluation of user solutions, while RxJS and WebSockets (Socket.io) power a real-time, interactive environment for seamless gameplay.

Real-time multiplayer (WebSockets)Pixel-level solution gradingAngular + Node.js + MongoDB
NodeJSExpress.jsAngularRxJSWebSocketsMongoDBGit

Education

2021 - Present

B.Sc. Computer Science (in progress)

RWTH Aachen University, Germany

Relevant Coursework — Data Structures and Algorithms, Software Engineering, Systems Programming Lab (C Programming) , Automated and Connected Driving Challenges (ROS-based perception, sensor fusion, planning, and V2X communication for autonomous vehicles)

Proseminar — Data Compression (Shannon's Noiseless Coding Theorem)

Seminar — Dominance-Based Evolutionary Multi-Objective Algorithms: NSGA-II and SMS-EMOA

2019 - 2021

German Language Course

Kaiserslautern University, Germany

Completed required german language classes up to level C1 and successfully passed the TestDaf C1 Examination

2017-2018

Computer Science, Networks & Telecommunications

E.S.G.I.S Private University, Togo

Fully funded scholarship in the field of computer science and networks for excellent grade on High School diploma

Classwork

01

Systems Programming Lab

Lab

RWTH Aachen University

Built OS-like components from scratch on a microcontroller, including a task scheduler and memory drivers. Implemented interrupt- and polling-based I/O, integrated peripherals such as an RFID reader, and worked with analog circuitry for A/D conversion.

Scheduler · Interrupts / Polling · I/O Binding · Memory Management · Peripheral Integration (RFID) · Analog Circuits (A/D Conversion)
C
02

Automated and Connected Driving Challenges

Course

Institute for Automotive Engineering (ika), RWTH Aachen University

Developed software for autonomous and connected vehicle systems within the ROS framework, covering sensor data processing, object fusion and tracking, vehicle guidance, V2X communication, and computer vision with neural networks.

Sensor Data Processing · Object Fusion & Tracking · Vehicle Guidance · V2X Communication · Computer Vision with Neural Networks
ROSPythonC++TensorFlowRVIZJupyterGit