Contributed to a McKinsey + QuantumBlack publication to develop a standardised methodology to identify, mitigate, and communicate risks across the full breadth of applied Machine Learning in industry. This work was presented at the ICLR 2019 Debug ML Workshop in New Orleans. Note, the authors are listed here in alphabetical order.
Co-led the mentoring team at MVA 2019, a competition which supports students building social enterprises in the UK&I. As part of my role, I introduced a series of webinars to connect students with experts in the field, including an entrepreneur from the Skoll Centre at Oxford, a social-impact VC from India, and Entrepreneur First.
Contributed to research which explored the application of Reinforcement Learning to discover better treatment policies for Sepsis. My contributions centred around investigating reward shaping and temporal dependencies. Our workshop paper was accepted to both the ML4H and TIML workshops at NIPS 2017.
Built Scholarly.ml - a tool to allow researchers to discover contemporaries who have published similar material and/or share similar research interests. The underlying algorithms used to categorise this data are entirely unsupervised. Try searching for "Yann Lecun" or "Generative Adversarial Networks"
Our team created SpatialRL - a software platform to facilitate training RL Agents through combining the Tensorflow interface with the ease-of-use of Unity and the scalibility of SpatialOS. We demonstrated that our approach could be used to train an RL agent using A3C, and were awarded the Improbable Prize at the event.
Created an algorithm to compress educational videos by 100x to reduce the data cost of accessing online education. Our team was awarded a prize at the Facebook Global Hackathon Finals. We productised and donated our algorithm to DotLearn, an MIT-based education startup who are trialling it in Ghana.
Currently working in London on data-driven analytics projects for blue-chip clients across a variety of advanced industries
• Led the design and development of a novel Cordova-based mobile transport app (“MotionMap”) to commercialise one of the world’s largest city-wide smart-sensor deployments in Milton Keynes, UK. • Developed internal web-tools to reduce the time spent manually annotating facilities within cities by more than five-fold. • Facilitated the installation of the sensor network by creating algorithms to help lower installation costs by > £50,000.
Masters in Information & Computer Engineering • 1st Year Grade: First Class • 2nd Year Grade: First Class • MIT GPA: 5.0 • Masters Grade: Distinction Masters Project: A smart stethoscope to automatically identify lung diseases using Machine Learning Activities: Cambridge University Eco-Racing (CUER) - Business Manager, Cambridge University Bangla Society - Treasurer, Hackbridge - Founder
• Developed an astronaut detection system for the International Space Station, using machine learning. This formed part of a NASA-led research project at MIT for an autonomous robot (“Astrobee”) which will be deployed on the ISS. • Improved an open-source detection system and implemented a multi-processing module for parallelised classification.
Cambridge-MIT exchange student with a concentration in Computer Science. • MIT GPA: 5.0
• Upgraded an outdated machine learning toolkit to give a 300% increase in training performance. • Designed a computer-vision algorithm in C++ to optimise a segmentation network by processing the model’s predictions and identifying uncertain output for model fine-tuning. This automated what was previously a time-intensive manual task. • Produced scripts to automate data retrieval from several sources for use in training machine learning algorithms.
• Led a 10-person team to raise funds for this student-led organisation with an operating budget in excess of £1m. • Overhauled the team’s former sponsorship structure and implemented a formalised system to facilitate fundraising. • Sourced funds to employ a team of full-time students which allowed us to improve the quality of our vehicle design.
• Developed high-level overviews of a safety-critical avionics system and identified areas for improvement; this was done to guide the design of a safer system. • Built software to automate the time-intensive manual process of cross-checking materials with EU-compliant standards.
A national prize from the Royal Academy of Engineering for young engineers who display entrepreneurial talent.
Awarded a scholarship for my performance in 2nd year Tripos Examinations at Cambridge.
A £5,000 scholarship from the Royal Academy of Engineering for students who demonstrate strong leadership potential.
Awarded a scholarship for my performance in 1st year Tripos Examinations at Cambridge.
Awarded a team prize for building the best robot in a competition for 1st year Cambridge Engineers.
Awarded an academic achievement prize by the Harvard Club UK.
© 2019 Imran Ahmed - Theme modified from 'Jalpc' by Jarekk
