Farid Ahmed

Developed a supervised machine learning workflow focusing on data preprocessing, model training, and evaluation for structured datasets. The work involved experimenting with different model configurations, analysing performance metrics, and documenting results.

The project strengthened practical ML engineering skills including feature engineering, debugging training behaviour.

Developed generative modelling experimentation pipelines for financial time-series simulation using PyTorch, improving reproducibility and evaluation consistency. Designed statistical validation workflows (VaR, ES, distribution metrics) to benchmark model performance and support research outputs. Collaborated on reproducible research practices.

Evaluated and validated large language model (LLM) outputs using structured review frameworks, identifying failure patterns and improving prompt alignment workflows. Contributed to QA processes by documenting edge cases, debugging model behaviour, and supporting safer deployment practices through detailed feedback and testing pipelines.

Built Python and SQL-based analytics pipelines to process operational datasets, developing regression and forecasting models to support data-driven decision making. Automated reporting workflows using Pandas and scripting, reducing manual analysis time and improving reporting efficiency across teams.

Supported backend product development by analysing AI platform workflows and translating technical requirements into API specifications and data-flow designs. Produced structured technical documentation that improved collaboration between engineering and product teams and clarified implementation priorities.

Activities and societies: Built end-to-end ML pipelines for preprocessing, model training, evaluation, and visualisation; implemented generative models for cryptocurrency return simulation and performance benchmarking.

MSc in Artificial Intelligence covering deep learning, NLP, scalable ML systems, and applied data science; dissertation on synthetic financial time-series generation using tail-aware GAN architectures with statistical risk evaluation.

Exchange semester at CY Tech University in Paris as part of undergraduate studies.

Activities and societies: Developed NLP classification pipelines using transformer embeddings and lightweight neural models; designed preprocessing workflows in Python to streamline experiments.

Bachelor of Technology in Computer Science with coursework in software engineering, algorithms, databases, machine learning, and distributed systems; final project on few-shot learning for LinkedIn experience classification.