Accelerating Algorithm Design


Accelerating Algorithm Design

Customers need results at the speed of their missions. Modern programming languages and concepts make it possible to develop many solutions on-demand, in response to operational needs on the timeframe of hours or days rather than months or years as in traditional development methodologies.

Impact-Driven Development (ID2)
A software development methodology that prioritizes testing against a mission use case as soon as possible, mitigating the risk of experimental techniques.
To provide next-generation analytic support, we created the Impact-Driven Development methodology: a high-speed, low-overhead technique that achieves proven mission results as fast as possible and then builds on that foundation. The method weaves design thinking into everyday programming, helping to minimize the risk of taking on even the most complex problems by trying multiple approaches very quickly. Engility methodologists have successfully demonstrated solutions in weeks or months for problems that others spent years trying to solve.

Driving Results

The average algorithm enables users to get to the results they want 100x-1000x faster than the best available manual process—if one even exists.
Accelerating Algorithm Design Engility teams develop dozens of algorithms each year to solve our customers’ most challenging problems using Engility’s Methodology Engineering and Impact-Driven Development. The average algorithm enables users to get to the results they want 100x-1000x faster than the best available manual process—if one even exists.

At the same time, our teams maintain a laser focus on getting to those outcomes, making smart planning tradeoffs to minimize project risk and create reliable solutions. In one notable case, an Engility team’s algorithm outperformed a competitor’s in under a quarter of the development time by “failing fastest”—finding ways to test on real mission data early so good methods can be identified quickly.

Engility recognizes that our customers need the right solutions, right away—and using our bottom-up, mission-centric algorithm design approach, we can reliably deliver those solutions against even the most challenging problem sets.

Modern computing is complicated, but here are a few useful definitions:

Data Analytics/Applied Statistics/Operations Research
Domain-specific models and corresponding statistical techniques used to identify and solve problems.
Machine Learning
General statistical techniques used to identify patterns in data for which a domain-specific model is unknown, unavailable or too complex to implement.
Artificial Intelligence
Techniques mirroring or mimicking an aspect of human thought, particularly including self-improvement with experience.

Big Data
Datasets so large that specific techniques must be applied to store, process and interpret them, even in the presence of powerful modern hardware.
Cloud Computing
Remote storage and processing allocated on demand rather than dedicated to one customer, limiting infrastructure costs.
Data Science
An interdisciplinary field combining elements of machine learning, data analytics, big data and cloud computing.

Share this Post:

Posted by Chris Milroy

I am an applied mathematician and Engility Technical Fellow for advanced analytics, machine learning, and artificial intelligence. I also lead the Advanced Data Solutions team and specialize in experimental adaptation of mathematical, physical, and social science models to solve hard data science problems. I am also a member of the IEEE and the ACM.