Quant Led Cloud Resource Automation: When options trading and cloud commitment management collide

At Archera, our primary goal is to enable frictionless innovation on the cloud through managing and optimizing cloud resources with our automated solution. Nikhil Khanna, Co-Founder and CTO of Archera, was integral towards building the platform. His unique quantitative background and experience in options trading has helped Archera build a truly next generation solution that leverages machine learning and complex financial modelling.

Before creating Archera, Nikhil worked as a Software Engineering Intern at D. E. Shaw in 2016, and as a Data Science Intern at Uber in 2017. Despite the difference in titles, he worked primarily on the financial side of both businesses.

At D. E. Shaw, Nikhil worked on the future’s trading team. He was responsible for examining the trade executions in a future’s market and diagnosing how the execution strategy and behavior changed in accordance with matching behavior in an exchange they were transacting through. His main goal was to evaluate the tradeoffs in choosing different execution strategies and therefore determining the best use-cases for each strategy.

At Uber, Nikhil worked in the financial data science field, spending most of his time combing through Uber’s financial data and searching for gaps in the data where money was not fully accounted for. In doing so, Nikhil worked with the Latin America team to diagnose a financial problem that they were having with their marketplace.

In many countries outside the US, Uber had a large cash business. This created an arrears (overdue payments) problem as Uber couldn’t efficiently track all the cash that drivers collected. Nikhil’s task was to diagnose the scope of the problem by looking at the current data at Uber, calculating the cost of the current arrears and forecasting what this number would grow to in the following few years. After discovering that the problem was quite large and warranted immediate action, Nikhil’s responsibilities shifted to working with the Latin American team to identify changes that could mitigate the issue, or at least track the cost of the problem in the short term to assist coming up with a long-term solution.

Nikhil enjoyed his experiences working at Uber and D. E. Shaw, and accomplished a lot during his time at both. However, working at these larger companies helped him realize that he was more interested in working in a smaller, cozier environment.

Nikhil said that as an intern he noticed, “a lot of the inefficiencies and dysfunctions that exist within a big company.” These inefficiencies could sometimes prevent those with great solutions from implementing them. This observation led him to realize that a company tasked with “addressing a net new problem” was something he was excited by.

Nikhil’s background in machine learning (ML), an extremely compute and data heavy process, introduced him to the net new problem. To get the necessary ML processing power and storage space, Nikhil would often look to cloud providers for a quick solution. However, when training models using machine learning, Nikhil noticed that his cloud bill for even simpler models was far higher than he had initially expected. This eventually led to him realizing “the bill [he] was paying to AWS was a more significant problem and a bigger point of stress for [him] than most of the problems [he] was trying to train models to solve”. Nikhil’s brother Aran (CEO, Co-Founder at Archera) was working at AWS at the time and noticed a similar disconnect between what users expected of their cloud bill and what it ended up being.

Working together, Nikhil and Aran created Archera with the vision of enabling frictionless innovation on the cloud. They wanted to take away the burden of continuously managing cloud resources from beginning to end and truly optimize purchasing to provide a balance between savings and engineering flexibility.

At both of his roles, Nikhil worked extensively on data analysis and financial forecasting, using Python and popular data science and machine learning tools to aid in his tasks. After finalizing his vision for Archera, Nikhil found that many of the skills he had already developed would be useful in actualizing this vision. At Archera, Nikhil is responsible for keeping Archera’s financial forecasting code up to date with the latest changes to AWS’ internal pricing logic.

He believes that what sets Archera apart from their competitors is that they try to “account for the fundamental uncertainty that exists within the cloud”. Archera does this by providing their customers with comprehensive payment plans that are structured around detailed queries of their expenditure till date alongside thorough forecasting of their likely future expenditure.

Nikhil said that it is essentially “a financial problem that [they] are trying to solve”. However, manually combing through all of the cloud expenditure for any customer, especially a larger one, is an impossible task. Thus, Archera’s solution combines automation with financial forecasting to streamline the cloud optimization process for their customers. Due to this, Nikhil’s background in machine learning and financial forecasting are invaluable assets to the company. His work at D. E. Shaw and Uber, where he performed similar analysis of financial data, has helped Archera become the company it is today.

In the short term, Nikhil’s biggest hope for Archera is for the UX to become more workflow oriented. He believes that while there is a lot of uncertainty surrounding financial forecasting of cloud usage, there is also “a lot of good information that’s out there but is not in a state that’s actionable right now”. Nikhil believes that in a typical organization, an engineer for a specific team might be able to predict the future state of that team’s section of the AWS infrastructure more accurately than perhaps a financial analyst could. However, this information is not actionable as it often remains as unspoken knowledge in that engineer’s head. Thus, Nikhil’s current aim is for the Archera solution to integrate with the typical workflow of a cloud-based company in order to provide a more accurate estimate of cloud-based usage needs.