Samuel Barber

MEng Electronic and Information Engineering - Predicted 1st Class Honours.

Relevant modules: Programming for Engineers, Software Systems, Instruction Architecture and Compilers, Discrete Mathematics, Information Processing, Advanced Computer Architecture, Machine Reasoning.

A-levels: Mathematics (A*), Computing (A*), Physics (A*), EPQ (A*).

Working on palm-based biometric authentication.

• Generating synthetic training data using diffusion models.
• Designed and operated large-scale distributed inference pipelines using AWS Spot instances.
• CNN performance profiling and optimisation.

Worked on palm-based biometric authentication.

• Designed and deployed a full-stack data collection system, gathering 2 TB of palm image data and 2,400 scans during the Edinburgh Fringe Festival using React, AWS S3, and Aurora.
• Developed a CNN-based region-of-interest detector for palm images, reducing the false-match rate by 30% and improving the false non-match rate by 20%.
• Fine-tuned an existing segmentation CNN for near-infrared imagery to improve performance.
• Implemented a scalable ML inference service enabling real-time, off-device model execution via RESTful APIs.

Collaborated with Renesas Electronics as part of a team to design an FPGA-based emulation system for power controller verification.

• Designed a modular FPGA-analog emulation platform to validate digital power controller architectures under real hardware conditions.
• Implemented remote monitoring and control for secure real-time observation and configuration of emulation runs.
• Utilised a PCIe interface for rapid reconfiguration of modular analog electronics, simplifying iteration and scalability.

Operated a business specialising in Bluetooth devices sold direct-to-consumer and wholesale.

• Sold over 20,000 items to more than 14,000 customers.
• Developed .NET-based tools to automate invoicing, emails, and postage labels.
• Independently funded university studies without taking a maintenance loan.

Breezy

breezy.mercurycomponents.co.uk

Portable air-quality sensing device and companion app that crowd-sources pollution data from commuters to suggest cleaner, lower-exposure routes and visualise live pollution heatmaps.

RISC-V CPU and C Compiler

Designed a pipelined RISC-V CPU with two-way cache and hazard detection using SystemVerilog, GTKWave and Verilator. Developed a C++ ANSI C compiler for RISC-V 32I supporting floats, doubles, unsigned types, functions, loops, pointers, and arrays.

FridgeHack (IC HACK 23) - Runner-up

AI web app to reduce food waste. Runner-up in the Create a Sustainable Future challenge and recognised as outstanding by Cisco.