Data at Santa Clara, California

Job Opportunity At Power

In March, 25% of all patients enrolling in clinical trials for depression came from Power. Help us get more promising new treatments into the market.

Work directly with a cofounder on a mix of data science & platform problems. Continue to gobble-up scope as you ship wins.

Own a number of key product surface areas (ranking, match, etc) that reliably drive large impact across the business.

Join a rapidly growing company (we've ~5x'd in the past year) while we're still just a tight-knit team of ~30.

Work with a product team where everyone writes their own queries, and as a result where Data's role is to provide them leverage (via better infra, models, etc etc).

Let's accelerate medical innovation. At Power, we're making clinical trials easy to navigate for all patients. Today, patients look to promising medical research when standard treatments have failed them. Unfortunately, there is no single source of truth that patients can understand and trust. The status quo is a government run website which is written in complex medical jargon that is impossible for most people to navigate.

At the same time, clinical trials are the bottleneck on medical innovation-new medicine takes almost 10 years to test and over 86% of all clinical trials are delayed because they can't find enough patients. This is one of the biggest problems facing the life science industry.

Power is the easiest way for patients to access promising new studies. Thus helping more patients find treatments and speeding up the progress of medical research as we go. Importantly, the platform really works:

• Our patients collectively schedule over X,000 screening appointments every month.
• Top clinical trial sites see a >50% lift across their business.
• Pharma companies running complex trials often receive >70%

The role. We're looking for someone to come be lead-up data at Power. For a long time, we got away without having a data function by ensuring all our PMs were sufficiently technical to do their own analysis.

As a result, many of the models powering our core systems were roughed in by me (Bask, a co-founder), in-between Zoom calls and over weekends. Lately, when I've revisited these systems, I've often been able to eke out an extra 10%+ lift in some key metric.

The embarrassing reality here is that it shouldn't be so easy to find wins of this magnitude. Given this, and our burgeoning scale, we're looking for someone to take over these key surface areas, and generally flex across all domains of data as is required by the business. You'd report directly to a cofounder, take ownership of our most-impactful decision systems on day one, and continue to gobble-up responsibility as we scale.

Some of the problems we'll be working on:

• Determine which patient a site should call first: We've just moved from an elastic net logistic regression to a CatBoost model predicting whether a patient will answer the phone. That said, we haven't done thoughtful feature-level interaction analysis, nor have we started to optimize against the full expected value that a patient will randomize for a trial.
• Predict how many patients we can enroll per month for new protocols or indications: How can we build a research-and-synthesis pipeline that takes in a protocol + site list, and spits out a feasibility estimate based on a mix of historical performance and published epi data?
• Determine how we should allocate marketing dollars across geographies: We've only recently implemented naive saturation of a) cost per patient as a function of paid spend by geo, and b) site call volume as a function of patients. I haven't at all started to consider that it's possible for us to ratchet up our filtering for the most cost-effective clinics.
• Determine which sites should have the chance to contact a given patient: Once a site gets in touch with a patient, we remove said patient from other sites' portals. This means that if Site A is way more likely to actually enroll a given patient than Site B, we might want to avoid showing the patient to Site B altogether. We need a model to drive policy here.
• Improve our internal view of patient eligibility: Today, the systems that tie-break across patient health profile data sources are all manually defined. In time, the prioritization here should be driven by statistical processes.

About you:

• You've spent a considerable part of your professional life analyzing data. You're probably intimately familiar with SQL and either Python or R.
• You're hungry for a role where the gap between the current state of the systems and what's possible is wide. You're excited, not nervous, about making calls about the stack.
• You have a track-record of driving business impact through both model development & analysis. You're comfortable working with ML-based approaches.
• You translate technical work into business decisions cleanly. Stakeholders leave your reviews with a clearer view of the world - not a longer reading list.
• You can drive projects independently or in a highly collaborative, x-functional manner. You naturally balance speed, craft, and business goals.

Learn more here - then send an email to View email address on click.appcast.io to chat.