Pluspunten
- Lark has figured out a way to make money and help people. This is exceptionally rare in health tech, and I can say this as someone who has spent their entire career in this industry. At my prior job where I grew a startup from the ground up, I worked with many, many health startups, including some "unicorns", and I have tracked theirs (and others') trajectories from D2C mission-driven products to ones that chiefly serve the interests of the paying customers - pharma, insurance companies, and device manufacturers - making patients a second-class citizen to the product and not really driving good patient outcomes. Health tech is hard because healthcare is expensive, and most people cannot pay for services outside of their regular doctor visits and prescriptions. It's very easy to build a new product in health tech. The hard parts are: (1) getting people to use it, (2) improving patient outcomes, and (3) making money. I do not know of a single health tech company out there that has succeeded at all three; most often, #2 and #3 are mutually exclusive. Lark excels at #2 and #3, and is steadily improving on #1. - Lark employees are some of the most collaborative and cooperative people I have ever worked with. Have a technical question? Someone will jump in and help you out without any judgment whatsoever. Having a rough day and need someone to listen to? A coworker will lend an ear and be there for you. I truly feel energized from people at work. - Fascinating technical challenges. I have never been bored for a second here at Lark. Providing a comprehensive health coaching platform means that we have to support many more features than a typical company of our size. We have a device data ingestion pipeline, a conversation engine, a device delivery and integration service, an eligibility rules framework, a streamlined data analytics pipeline, a real-time streaming event platform, and all of these pieces of infrastructure present unique challenges. The conversation engine itself presents perhaps the most fascinating technical problem -- how do you ensure that the latest coaching conversation the chatbot provides is reflective of the most recent user data? We can't rely on naive asynchronous processing -- we need truly real-time data processing. I've seen other companies tackle this problem, but they have solely been focused on telemetry and surfacing data to users in reports. Lark uses this data for live, automated coaching, which makes the problem space much more interesting.
Minpunten
- Not enough go-getters and proactive owners. Sometimes with a culture of people that are so easygoing and collaborative, it's hard to get people to push things forward quickly. It's always a small subset of people who tend to speak up and take action. We need to achieve more of a balance. - Legacy infrastructure is frustrating to work with and lacks documentation. We need to continue setting a high bar for our new infrastructure and, when necessary, fill in the gaps in our legacy code rather than wait for it to be deprecated. - Clearer guidance and coaching. We have a pretty good idea of where we want to go - we just need more of a game plan on how specifically to get there. We may have to take a winding path, but it would be helpful for leadership to provide clear markers along the way, and a map, rather than simply a north star. - Platform-based teams. In order to scale as we grow, we need to move to having cross-functional teams, so that we don't have communication mishaps or delays between front-end, back-end, and devops orgs.