Code Sprints Case Studies
For MVPs, new features, and concept validation
Automating Advanced Manufacturing Strategy at Elkem
Industry Partner: Chris York VP from Elkem (Metal Production)
Story by Raúl: With the promising future of electric cars, battery production became the most concerned topic in 2021. Working with Elkem, the problem was to optimize and automate the inspection process in order to implement for crucibles. Those crucibles are needed to create graphite parts of lithium batteries for electric vehicles, which is a core part of the battery in itself. The team employed technologies like machine learning, Lean Six Sigma and some business optimization methods, and softwares like TIA Automation Portal, Abaqus and CATIA to design this system.
This optimization would mean a great advance for the EV industry and the ecological economy. By the end of 2 weeks, the team made sure that the Elkem engineering team could directly start the implementation process.
Team members: Amynaz, Sealani, Trevaughn Morrison, Raúl Sotillo
Predicts Buildings Structure, an AI Project with Deep Excavation LLC
Industry Partner: Dimitrios Konstantakos, CEO of Deep Excavation LLC
Story by Muhammad:
The team decided to address a critical pain point for stakeholders in urban infrastructure decision-making. The cost of projects can rise in the billions of USD because the decisions for "alignment selection" lacks predictability at the early stage. They focused on providing a proof of concept on a small subset of the overall solution to predict the structural elements of an individual building floor.
They quickly realised how manually laborious this task was going to be, covering four co-dependencies. (1) To locating existing publicly available data and code (in their limitations), (2) to adapting (to these limitations) by leveraging digital tools and specialist knowledge, (3) while assembling, and writing code to mimic outputs of manual effort or desired output (supervised machine learning), (4) and then to validate and graphically present.
The killer application remained consistent whereas the "pathway" reformulated relative to pivots and refinement (i.e. how we solve the problem), in order to ensure alignment of the aforementioned co-dependencies and high precision so, to acquire a strong, reliable foundation for further work. To everyone's delight, success was met in delivering the proof-of-concept for a scalable ground-breaking tool capable of predicting structural systems of existing buildings.
Members: Pablo, Hannah, and Muhammad
Precision Improvement Agriculture through precision with ML/AI with Applied Materials
Industry Partner: Sushil D. Padiyar, Senior Director at Applied Materials
Story by Samir and Nicquary : Pre-existing research shows that spectral imaging analysis detects disease and stress symptoms in plants before physical signs starting to show. The need of development of low cost on chip spectral camera modules generated our decision to focus on incorporating such a sensor in a IOT device. The MVP of Precision is that agriculture accounts for 70% of worldwide water use, so water usage must be optimized first, while incorporating hyperspectral imaging and SiVM (simple volume maximisation) to identify hydration stress in plants while the model determines a suitable water cycle and signs to make which adjustments.The biggest salvation for the team was changing the scale of the project to suit greenhouse farmers and those local to the US with more capital to test the project and gather data as well as efficacy. They used a CAD design software to create an engineering model and a graphical flow chart tool to create a ML functional flow diagram as well as Google's Jamboard to create the task list for the software structure.
Members: Samir, Babak, Kymani, and Nicquary