Departments / AI Research
EduRankAI is a frontier AI research lab building reasoning systems and foundational models from the ground up — not fine-tuning what already exists, but training new architectures that learn the way the most curious humans do. We are hiring researchers, research engineers, and research scientists who want to work on real problems in reasoning, alignment, evaluation, and learning dynamics. This is research-led work — publications, internal papers, and reproducible experiments are part of the job. Selection here is genuinely competitive. The application fee reflects that — every shortlisted candidate gets a personal review by the research team.
Build foundational AI from first principles — proprietary language models, agentic systems, training infrastructure and the compilers and runtimes beneath them.
Python · PyTorch · Transformer architectures · Deep learning + 48 more
Build production computer-vision systems — detection, pose, video analytics, 3D vision — that run in real time.
Python · C++ · PyTorch · OpenCV + 8 more
Running AI models directly on constrained devices — a real optimisation discipline connecting our AI research to embedded hardware.
Model quantisation/optimisation · TensorFlow Lite/ONNX · Embedded basics · Python + 1 more
A research-track internship for students aiming at genuine scientific contribution — designing experiments, forming hypotheses, and writing up results to the standard of a workshop paper.
Python · Research methodology · Experimental design · Academic writing + 2 more
The infrastructure behind AI research — training pipelines, experiment tracking, compute scheduling — determines how fast the research team can actually iterate.
Python · ML pipeline tools (MLflow/Weights & Biases or equivalent) · Docker basics · Cloud/GPU compute basics + 1 more
Responsible AI — bias, fairness, transparency, and accountability in the models we ship to learners — is a practical governance discipline that complements our safety and alignment research.
Responsible AI frameworks · Bias/fairness evaluation basics · Python · Documentation + 1 more
Alignment research — making model behaviour match intended goals, especially in a tutoring context where "helpful" must also mean "pedagogically honest" — is a distinct research thread from general safety.
Python · Alignment techniques (RLHF basics) · Preference-learning concepts · Research literature review + 1 more
The engineering side of large language models — serving, optimising, and integrating them reliably — distinct from the research side; for students who want to build production LLM systems.
Python · LLM APIs (OpenAI-compatible/self-hosted) · Inference optimisation basics · API integration + 1 more
Multimodal models — reasoning across text, image, audio, and video together — are directly relevant to how AquinTutor teaches with diagrams, voice, and visual explanations.
Python · PyTorch/TensorFlow · Multimodal model architectures · Data preprocessing (image/audio/text) + 1 more
Agentic systems — models that plan, call tools, and take multi-step actions — underlie our roadmap for autonomous tutoring workflows and internal automation.
Python · Agent frameworks (LangChain/LlamaIndex or equivalent) · Tool-use/function-calling patterns · Evaluation design + 1 more
Build the evaluation harnesses that keep our AI systems honest — benchmarks, regression tests, and bias/safety checks before anything ships.
Python · Testing methodology · Statistics basics · Attention to detail + 1 more
Design, test, and systematise the prompts that power our AI tutor, content generation, and evaluation systems — a discipline that blends writing craft with empirical rigour.
Strong written communication · LLM behaviour understanding · Basic Python · Structured testing mindset