We’re growing! Why join Premise?

By Giri Anantharaman, Principal Architect

Transporting goods in Sekondi-Takoradi, Ghana, as taken by a Premise contributor.

I may be biased, but I truly believe that Premise is tackling some of today’s most fascinating engineering challenges. Why? Because we’re dealing with the inherent unknowns and uncertainty of making sense of how the real world is evolving — and the real world is hard.

To put this into concrete terms, we’re in the Philippines measuring and monitoring the impact of a tax law on cigarettes. In Brazil we’re building inflation indicators. In East Africa we’ve mapped financial access and electrification patterns. In other words, we’re not building for the sake of building — we’re building to materially and positively impact the way people make decisions.

To help us make even faster headway on our roadmap… we’re hiring! Here’s a quick look at some of the biggest engineering challenges we’re working on right now:

  • Task allocation: We’re orchestrating thousands of people in countries around the world to collect highly structured data using Android smartphones. Our platform allocates tasks based on user preferences and capabilities while constrained by cost (we pay our contributors), the nature and complexity of tasks, and desired data coverage and quality. The hardest part is figuring out how to match tasks to users, ensuring we get optimal coverage within budget. The tasks are diverse — from collecting petrol prices and black market exchange rates to documenting hygiene practices and the quality of rural, unmapped roads. It’s truly a complex recommender system problem.
  • Survey design: Our tools are capable of creating a wide range of surveys to quantify a broad set of problems based on several underlying data taxonomies. Therefore our underlying data models must represent a wide range of data inputs, which means a complex UX challenge. Now, imagine a set of simple inputs such as text or numbers, and marrying those with complex geospatial constraints such as guiding a contributor in the Philippines down a half-paved rural road that isn’t on an official map — and all in a single web interface.
  • App design: We focus exclusively on Android, the dominant platform in the markets where we have the largest footprint. Our contributors are using a diversity of device types, which means unique UX and implementation constraints, in addition to variable connectivity constraints across the world. We can’t count on Google Play Services given our device footprint, which forces us to get creative about complex survey delivery localized in a dozen languages.
  • Data science and machine learning: We’re applying techniques from recommender systems to deep convolutional nets to build a system ultimately intended to strengthen economies and advance human progress. Because we’re gathering such a wide range of data, our QA process is currently based on a mix of integrity checks on inputs, image recognition problems like object detection (is it a tomato or a carrot, and are those carrots loose or packed in bunches?), text extraction from images, and complex geospatial analysis on geotagged submissions. One of the biggest challenges is determining how to balance between automated versus manual checks for data quality since our surveys and contributors constantly evolve. We need to combine the best of both in machine learning or statistical inference to make this scalable.
Market bounty in Bamako, Mali (left); Stalls in Rubavu, Rwanda (right).
  • Payments galore: We’re trying to bring some semblance of elegance to the inherently messy global payments landscape. To pay our contributors, we currently have more than a dozen payments providers — from Coinbase to PayPal to AstroPay — integrated into our app. Next up is figuring out how to unify and scale these payment mechanisms.

In the coming year, we’ll be focused on our mobile apps, scaling our data model, survey management tools, data analysis and task allocation algorithms.

If you’re a data, Android, web front-end, back-end or machine learning engineer, value a passionate, tight-knit, super-driven team that moves fast, appreciates candor, and deeply values the diversity of our backgrounds, please visit our careers page and say hi.