Job Title: AI/ML Engineering Intern (Generative AI, Agentic Systems & Cloud)
machine learning, deep learning, Generative AI, AI agents and multi-agent systems
You will work closely with our AI engineering team to understand business problems, experiment with approaches, build working solutions and support their deployment. Were looking for candidates who can write code, explain their technical decisions and systematically investigate failures.
In this role, you will:
- Develop and evaluate machine learning and deep learning models for classification, prediction, natural language processing and other business use cases.
- Prepare datasets, perform exploratory data analysis, engineer features and build reproducible training and evaluation pipelines.
- Build
LLM-powered applications and AI agents
that retrieve information, interact with tools and carry out defined tasks. - Develop
agentic workflows
involving task decomposition, planning, tool selection, routing, state management and human approval where appropriate. - Implement
multi-agent orchestration
, including specialised agent roles, supervisor–worker patterns, delegation, agent handoffs and sequential or parallel execution. - Build
retrieval-augmented generation (RAG)
pipelines covering document ingestion, chunking, embeddings, vector search, retrieval and reranking. - Integrate models and agents with APIs, databases and external services using structured outputs and function/tool calling.
- Evaluate model and agent behaviour, investigate errors and improve accuracy, reliability, response latency and inference cost.
- Support model serving and cloud deployment, and write maintainable code, automated tests and technical documentation.
- Collaborate through Git, pull requests and reproducible experiments.
You should bring the following knowledge and skills:
Python programming:
Strong understanding of data structures, functions, classes, object-oriented programming, exception handling, debugging, file processing and package management. Familiarity with asynchronous programming, concurrency and virtual environments.Data preparation and analysis:
Practical experience with NumPy and pandas
, including missing-value handling, outlier analysis, categorical encoding, scaling, feature engineering and dataset preparation. Ability to explore and communicate findings using Matplotlib or Seaborn
.Machine learning fundamentals:
Supervised and unsupervised learning, classification, regression, clustering and dimensionality reduction. Familiarity with approaches such as linear/logistic regression, decision trees, random forests, gradient boosting, k-means and PCA, using scikit-learn
or equivalent libraries.Model training and evaluation:
Train/validation/test splits, cross-validation, hyperparameter tuning, class imbalance, overfitting, underfitting, regularisation and data leakage. Ability to select appropriate metrics, including precision, recall, F1, ROC-AUC, MAE and RMSE
, and interpret model errors rather than relying on a single score.Deep learning:
Neural networks, activation functions, loss functions, backpropagation, optimisers, learning rates, dropout and transfer learning. Understanding of CNNs, RNNs/LSTMs and Transformers
, with hands-on experience in PyTorch or TensorFlow
.NLP and Generative AI:
Text preprocessing, tokenisation, embeddings, attention, language models and LLM inference. Familiarity with Hugging Face Transformers or LLM APIs
, prompt engineering, few-shot prompting, structured outputs, context-window management, grounding and hallucination handling.RAG and semantic retrieval:
Document processing, chunking strategies, embedding models, similarity search, metadata filtering, hybrid retrieval and reranking. Practical familiarity with vector-search tools such as FAISS, pgvector or Pinecone
, and an understanding of retrieval quality versus answer quality.Agentic AI:
Working understanding of agent execution loops, planning, tool calling, conversation state, memory, retries, fallbacks, termination conditions and human-in-the-loop controls. Ability to recognise when a deterministic workflow is more appropriate than autonomous agent behaviour.Multi-agent systems:
Understanding of agent coordination, delegation, communication, shared context and failure handling. Practical exposure to LangGraph, LangChain, CrewAI, AutoGen or equivalent frameworks
is valuable; familiarity with every framework is not required.Backend and data integration:
REST APIs, HTTP, JSON, input validation and API development using FastAPI or Flask
. Basic SQL
, database operations and familiarity with relational or NoSQL storage.Cloud and infrastructure:
Familiarity with AWS, Azure or Google Cloud
, including compute, object storage, IAM/access permissions, serverless functions, environment configuration and logging. Exposure to services such as Amazon S3, Lambda, EC2, Bedrock or SageMaker
, or their equivalents, is useful.Engineering and MLOps fundamentals:
Git, Linux, Docker and CI/CD
, along with experiment tracking, model/data versioning, reproducibility, model serving and monitoring. Awareness of model drift, deployment failures, latency and resource costs.
fine-tuning, LoRA/PEFT, Model Context Protocol (MCP), agent tracing, MLflow, automated AI evaluation or inference optimisation
Recent graduates in Computer Science, AI/ML, Data Science or related disciplines are welcome. Strong practical projects and demonstrable technical skills are essential.