Joel Jennings, Machine Learning Developer in Cambridge, United Kingdom
Joel Jennings

Machine Learning Developer in Cambridge, United Kingdom

Member since December 20, 2019
Joel has a strong mathematics background and spent three years working as a software engineer at a consultancy firm, specializing in embedded software while gaining a broad knowledge of different technologies. He has worked as a machine learning engineer on a range of tasks, from implementing algorithms in research papers to engineering deployed machine learning systems. He is driven by projects that use exciting theories to solve real-world problems.
Joel is now available for hire




Cambridge, United Kingdom



Preferred Environment

Emacs, Linux, Git

The most amazing...

...project I've worked on was using machine learning to decode neural signals being sent down the vagus nerve of a live subject.


  • Machine Learning and Software Engineer

    2020 - PRESENT
    • Published a multi-agent reinforcement learning paper at ICML with Huawei.
    • Deployed TensorFlow audio models to a Raspberry Pi using TensorFlow Lite. The audio was detected using a directional microphone and would allow the attached camera to focus on areas with anomalous sounds.
    • Developed a demonstrator of the UK highways traffic data that provides real-time information about the states of the roads to enable operators to more quickly detect and respond to congestions and accidents.
    • Created a Python Flask back end running on AWS to allow shoppers to offset their products' carbon through Shopify.
    Technologies: Python, PyTorch
  • Machine Learning Research Lead

    2020 - 2021
    • Developed a pipeline for analyzing peripheral nervous system data.
    • Created an iOS Watch app for assessing patients' ability to perform a six-minute walk test.
    • Managed a team of 6 ML researchers, engineers, and neural scientists.
    Technologies: Machine Learning, PyTorch
  • Machine Learning Team Lead

    2017 - 2020
    • Created a time series modeling library in TensorFlow for developing Gaussian Process models with faster inference through stochastic differential equation techniques.
    • Took responsibility for taking Gaussian process models from research code into a finance product using Kubernetes, Docker, and Airflow.
    • Implemented neural network-based algorithms in TensorFlow that led to paper publications at machine learning conferences as part of the multi-agent reinforcement learning team.
    • Served as the technical lead for a logistics time series forecasting customer project.
    • Managed the professional development of several machine learning engineers and performed interviews and coding test reviews as part of the recruitment process.
    Technologies: Statistics, Python, TensorFlow, PyTorch
  • Senior Embedded Software Engineer

    2013 - 2016
    Cambridge Consultants
    • Developed a satellite remote sensing camera for the Zoological Society of London to study animals and prevent poaching, using an Atmel AVR and Raspberry Pi. The camera was deployed in Kenya and Antarctica.
    • Created a virtual queuing wristband for theme parks involving NFC, Bluetooth, and LoRa on a Nordic nRF52 (based on a ARM Cortex M4). See Portfolio Projects for more information.
    • Created iPad app for asset tracking for the placement of US internet cables for a telecommunications company.
    • Developed a Bluetooth Low Energy EpiPen iPhone app, that guided users through how to perform an injection.
    • Developed the firmware of a demonstrator of an energy harvested Bluetooth enabled insulin injector. This device was capable of communicating dosage amount to a smartphone, using only the energy from physically injecting the insulin.
    • Wrote a tool to automatically generate C code for Bluetooth LE profiles for a semiconductor company.
    • Developed Bluetooth audio applications using CSR hardware (now part of Qualcomm).
    Technologies: Embedded C, C, Consulting


  • Mean Field Games

    In research published at a top AI conference, I was responsible for performing experiments and validation for using Reinforcement Learning to play games that involved thousands of agents. This involved creating and training neural networks in TensorFlow. The techniques developed in the paper could be used for placing mobile network points for major public gatherings.

  • Underground Asset Tracking

    Underground asset tracking is a huge issue when it comes to such things as digging in the ground and trying to avoid hitting existing utilities. I worked on a universal mobile app that allows engineers to query and locate any underground cabling. The app initially targeted the iPad but was built with technologies that allowed for its subsequent port over to Android. The Titanium framework was used together with Java, Objective-C, and JavaScript to bring together all the essentials in connecting an iPad to a Bluetooth device and data visualizations for the assets via the Google Maps API.

  • Virtual Queueing Wristband for a Theme Park

    I was part of a team that developed a wristband to create a virtual queuing waterpark in Florida. This device had Bluetooth, LoRa, and NFC radios. It had to have tiny power consumption to allow it to last all day and also had to be waterproof. It used a Nordic nRF52 SoC (based on an ARM Cortex M4).

    I wrote the bootloader to allow software updates over Bluetooth, implemented the framework, and the drivers for the LED display, as well as parts of the NFC interface that were used to allow the wristband to make payments and initiate queueing. I was responsible for optimizing power consumption.


  • Languages

    Python, C, Embedded C, Python 2, Python 3, C++, Haskell, Rust, Java, Objective-C, Idris, TypeScript, Swift
  • Libraries/APIs

    TensorFlow, NumPy, PyTorch, Matplotlib, Pandas, LAPACK
  • Platforms

    Raspberry Pi, ARM Linux, Linux, Bluetooth LE, Amazon Web Services (AWS), Android, MacOS, Windows
  • Other

    Artificial Intelligence (AI), Embedded Systems, Statistics, Algorithms, Machine Learning, Neural Networks, Sensor Data, Sensor Fusion, Hardware Drivers, Embedded Hardware, Bluetooth, Probabilistic Graphical Models, Bayesian Inference & Modeling, Bayesian Statistics, Deep Neural Networks, Reinforcement Learning, Deep Reinforcement Learning, Linear Algebra, Kalman Filtering, Hardware, Near-field Communication (NFC), LoRa, Consulting, People Management, Time Series, Mathematics
  • Tools

    Emacs, Vim Text Editor, Git, GitHub, Bitbucket, Subversion (SVN)
  • Paradigms

    Functional Programming, Object-oriented Programming (OOP), Agile, Management
  • Frameworks

    Flask, Boost


  • Ph.D. in BioPhysics
    2010 - 2013
    University of Cambridge - Cambridge, United Kingdom
  • Master's Degree in Computational Biology
    2009 - 2010
    University of Cambridge - Cambridge, United Kingdom
  • Bachelor's Degree in Management Studies
    2008 - 2009
    University of Cambridge - Cambridge, United Kingdom
  • Bachelor's Degree in Mathematics
    2005 - 2008
    University of Cambridge - Cambridge, United Kingdom

To view more profiles

Join Toptal
Share it with others