I DoorDash about 5 months in the past. That is my first time beginning at a brand new firm as a Information Science Supervisor. DoorDash strikes quick, expectations are excessive, and the area context is deep, which makes onboarding difficult. Nonetheless, it has additionally been one of many fastest-growing durations of my profession.
The primary three months at any new job are basically a constructing part — constructing connections, area understanding, and information information — and a clean onboarding units the inspiration for later success. Due to this fact, on this article, I’ll share what mattered most to start with months and my guidelines for any information science onboarding.
I. Construct Connections
Earlier than anything, let me begin with constructing connections. Once I was in school, I pictured information scientists as folks spending all day lengthy heads-down writing code and constructing fashions. Nonetheless, as I grew to become extra senior, I spotted that information scientists make actual impacts by embedding themselves deeply within the enterprise, utilizing information to determine alternatives, and driving enterprise choices. That is very true at present with tighter DS headcount and AI automating primary coding and evaluation workflows.
Due to this fact, constructing connections and incomes a seat on the desk must be a high precedence throughout onboarding. This consists of:
- Frequent onboarding periods along with your supervisor and onboarding buddy. These are the individuals who greatest perceive your future scope, expectations, and priorities. In my case, my supervisor was my onboarding buddy, and we met nearly day by day throughout the first two weeks. I at all times got here with a ready checklist of questions I encountered throughout onboarding.
- Arrange meet-and-greet calls with cross-functional companions. Right here is the agenda I often comply with in these calls:
- 1. Private introductions
- 2. Their focus space and high priorities
- 3. How my staff can greatest assist them
- 4. Any onboarding recommendation or “things I should know”
- I particularly just like the final query because it persistently gives nice insights. 5 years in the past, after I onboarded at Brex, I requested the identical query and summarised the responses into classes right here. The perfect I received this time is “Don’t be afraid to ask dumb questions. Play the new-hire card as much as possible in the first three months.”
- For these key companions, arrange weekly/bi-weekly 1:1s and get your self added to recurring challenge conferences. You could not contribute a lot at first, however simply listening in and gathering the context and questions is useful.
- In case you are onboarding as a supervisor like me, you need to begin speaking to your direct stories early. Throughout onboarding, I intention to study three issues from my direct stories: 1. Their tasks and challenges, 2. Their expectation of me as a supervisor, 3. Their profession objectives. The primary helps me ramp up on the realm. The latter two are essential for establishing belief and a collaborative working relationship early on.
II. Construct Area Context
Information scientists succeed once they perceive the enterprise effectively sufficient to affect choices — not simply analyze outcomes. Due to this fact, one other precedence throughout onboarding is to construct your area information. Frequent methods embody speaking to folks, studying docs, looking Slack, and asking loads of questions.
I often begin with conversations to determine key enterprise context and tasks. Then I dig into related docs in Google Drive or Confluence, and browse Slack messages in challenge channels. I additionally compile the questions after studying the docs, and ask them in 1:1s.
Nonetheless, one problem I bumped into is digging into the rabbit gap of docs. Every doc results in extra paperwork with quite a few unfamiliar metrics, acronym names, and tasks. That is particularly difficult as a supervisor — if every of your staff members has 3 tasks, then 5 folks means 15 tasks to catch up. At one level, my browser’s “To Read” tab group had over 30 tabs open.
Fortunately, AI instruments are right here to rescue. Whereas studying all of the docs one after the other is useful to get an in depth understanding, AI instruments are nice to offer a holistic view and join the dots. For instance,
- At DoorDash, Glean has entry to inside docs and Slack. I usually chat with Glean, asking questions like “How is GOV calculated?”, “Provide a summary of the project X, including the goal, timeline, findings, and conclusion.” It hyperlinks to the doc sources, so I can nonetheless dive deeper shortly if wanted.
- One other instrument I attempted is NotebookLM. I shared the docs on a selected matter with it, and requested it to generate summaries and thoughts maps for me to gather my ideas in a extra organized approach. It will possibly additionally create podcasts, that are generally extra digestible than studying docs.
- Different AI instruments like ChatGPT may also connect with inside docs and serve an identical goal.
III. Construct Information Information
Constructing information information is as essential as constructing area information for information scientists. As a front-line supervisor, I maintain myself to a easy normal: I ought to be capable of do hands-on information work effectively sufficient to offer sensible, credible steering to my staff.
Here’s what helped me ramp up shortly:
- Arrange tech stack in week one: I like to recommend organising the tech stack and developer setting early. Why? Entry points, permissions, and peculiar setting issues at all times take longer than anticipated. The sooner you might have the whole lot arrange, the earlier you can begin taking part in with the information.
- Make full use of AI-assisted information instruments: Each tech firm is integrating AI into its information workflows. For instance, at DoorDash, we’ve got Cursor linked to Snowflake with inside information information and context to generate SQL queries and evaluation grounded in our information. Although the generated queries usually are not but 100% correct, the tables, joins, and previous queries it factors me to function glorious beginning factors. It gained’t change your technical judgment, however it dramatically reduces the time to first perception.
- Perceive key metrics and their relationships: Information information not solely means with the ability to entry and question the information, however perceive the enterprise from an information lens. I often begin with weekly enterprise critiques to seek out the core metrics and their development. That is additionally an effective way to contextualize the metrics and have an thought of what “normal” seems like. I’ve discovered this extremely useful when gut-checking analyses and experiment outcomes later.
- Get your arms soiled: Nothing enforces your information understanding greater than doing a little hands-on work. An excellent onboarding program often features a mini starter challenge. Whilst a supervisor, I did some IC work throughout my onboarding, together with alternative sizing for the planning cycle, designing and analyzing a number of experiments, and diagnosing and forecasting metrics motion. These tasks accelerated my studying way over passive studying.
IV. Begin Small and Contribute Early
Whereas onboarding is primarily about studying, I strongly advocate beginning small and contributing early. Early contributions sign possession and construct belief — usually quicker than ready for a “perfect” challenge. Listed here are some concrete methods:
- Enhance the onboarding documentation: As you undergo the onboarding doc, you’ll run into random technical points, discover damaged hyperlinks, or discover outdated directions. Not simply overcoming them your self, however enhancing the onboarding doc is an effective way to indicate that you’re a staff participant and need to make onboarding higher for future hires.
- Construct documentation: No firm has excellent documentation — from my very own expertise and chatting with my associates, most information groups face the problem of outdated or lacking documentation. As you might be onboarding and never busy with tasks but, it’s the excellent time to assist fill in these gaps. For instance, I constructed a challenge listing for my staff to centralize previous and ongoing tasks with key findings and clear factors of contact. I additionally created a set of metrics heuristics, summarising the causal relationship between completely different metrics we discovered from previous experiments and analyses. Word that every one these paperwork additionally develop into beneficial context for AI brokers, enhancing the standard and relevance of AI-generated outputs.
- Counsel course of enhancements: Each information staff operates in another way, with professionals and cons. Becoming a member of a brand new staff means you deliver a recent perspective on staff processes and may spot alternatives to enhance effectivity. Considerate ideas primarily based in your previous expertise are tremendous beneficial.
In my view, a profitable onboarding goals to ascertain cross-functional alignment, enterprise fluency, and information instinct.
Right here is my onboarding guidelines:
- Week 1–2: Foundations
– Meet key enterprise companions
– Get your self added to core cross-functional conferences
– Perceive staff focus and priorities at a high-level
– Arrange tech stack, entry, and permissions
– Write your first line of code
– Learn documentation and ask questions - Week 2–6: Get your arms soiled
– Deep dive into staff OKR and generally used information tables
– Deep dive into your focus space (extra docs and questions)
– Full a starter challenge end-to-end
– Make early contributions: Replace outdated data, construct one piece of documentation, or counsel one course of enchancment, and so on. - Week 6–12: Possession
– Have the ability to converse up in cross-functional conferences and supply your data-informed perspective
– Construct belief because the “go-to” individual to your area
Onboarding seems completely different throughout firms, roles, and seniority ranges. However the ideas keep constant. In the event you’re beginning a brand new position quickly, I hope this guidelines helps you ramp up with extra readability and confidence.



