The impact is staggering: nearly 70% of organizations report that technical debt significantly impairs their ability to innovate. Every unused dashboard and dusty model creates drag, making every new release harder and slower.

The Solution: Embrace the “Deprecation-First” Rule

Here’s the game-changing principle that successful data teams are using: To build something new, you must retire something old.

The rule is elegantly simple:

• New dashboard → deprecate one

• New metric → deprecate one

• New pipeline or model → same rule applies

Start with a 1:1 ratio, then scale to 1:3 or 1:5 as your team builds the muscle and focus for systematic cleanup.

How to Make It Work

Successful implementation requires four key elements:

1. Establish the Baseline Rule
Make it non-negotiable: no new asset gets built without retiring an old one.

2. Track Your Progress
Count additions versus deletions in every sprint. This simple metric keeps the team accountable and shows progress over time.

3. Protect Cleanup Time
Dedicate approximately 15% of your sprint capacity specifically to debt cleanup. This isn’t overhead; it’s an investment in future velocity.

4. Celebrate Success
Give team recognition for retiring dashboards, deprecating metrics, and cleaning up technical debt. What gets celebrated gets repeated.

Why This Matters More Than Ever

Technical debt isn’t just a maintenance issue; it’s an innovation killer. Every piece of unused infrastructure steals budget and slows down your team’s ability to deliver value. By systematically reducing complexity while building new capabilities, you create a virtuous cycle of increased agility and reduced maintenance overhead.

Ready to Get Started?

The beauty of this approach is that it requires no massive overhaul. Simply pair your next new build with something you can clean up. Ask your team: what could we deprecate this week?

The path to managing technical debt isn’t about perfect planning or comprehensive audits. It’s about building sustainable habits that prevent debt from accumulating faster than you can pay it down. Start small, stay consistent, and watch as your team’s velocity and innovation capacity begin to improve.

When I took over data at Opendoor 3 years ago, I walked into a classic problem: a lot of smart, junior people spread way too thin. Everyone was solving the same problems in their own little corners. It was organized chaos, with duplicated work and no single source of truth.

My first move wasn’t to hire more people. It was to hire different people. I stopped hiring pods of junior analysts and instead brought on senior IC’s for each major business area. We centralized the team but kept them embedded in the business. The signal instantly got clearer than the noise. We went from fighting fires to shaping strategy.

Fast forward to now: Hiring senior IC’s isn’t enough. We are in the Super IC era.

Forget 10x. We’re in the 100x Era Now.

We used to get excited about the mythical “10x engineer.” That’s ancient history. In 2025, with AI in everyone’s toolkit, the best people aren’t 10x. They’re 100x. One great data scientist can now do the work of an entire team.

The gap between an average contributor and a superstar has become a canyon. This isn’t just about being a faster coder. It’s about strategic thinking. It’s the difference between someone who can answer your question, and someone who tells you the question you should be asking.

Stop Taking Orders, Start Driving Outcomes.

This is why the old “service desk” model for data teams is so broken. If your team is just sitting around waiting for tickets, you’ve already lost. You’re a support function, not an engine for growth.

I pushed my team at Opendoor to think like product owners. Find the problem, build the solution, own the outcome. Don’t just deliver a report; build a data product that makes the report unnecessary. If you’re an IC, your job is to be proactive. Find the opportunity before your manager even knows it exists.

Your Best Ideas Now Have an Expiration Date.

Here’s another hard truth: speed is everything. With AI, that brilliant insight you have today could be an automated feature for your competitor next quarter. You can’t afford to wait for permission or polish a proposal for weeks.

Build the ugly prototype. Get it in front of people. Show, don’t just tell. A good idea launched this week is infinitely better than a perfect one you’re still planning next year.

What to Do If You’re Not at a FAANG Company

Of course, this has created a crazy talent war. You see headlines about multi-million dollar packages for AI researchers, and it’s not just hype. One person really can change the game, and companies are paying for it.

So what can you do if you’re at a more traditional company? Act like a super IC anyway.

• Be a self-starter. Don’t wait for permission to fix something that’s broken.

• Make AI your superpower. Automate the boring stuff so you can focus on what matters.

• Think like the business. Connect every line of code to a customer problem or a dollar sign.

• Learn out loud. Share what you’re working on online or internally. You’ll be surprised who’s listening.

For leaders, the job is to find these people, pay them what they’re worth (and yes you may have to battle with your CFO about this!), and give them the two things they crave most: autonomy and interesting problems to solve.

Ultimately, this isn’t about humans vs. AI. It’s about a new kind of team: talented people using incredibly powerful tools. The future isn’t scary; it’s an upgrade. Don’t be afraid of the super IC. Figure out how to become one.

(Note: written with the help of AI tools for editing)

Great candidates don’t just talk about the project they’re most excited about. They talk about the project I (as the interviewer) care about.

Here’s what I mean.

Years ago, before Microsoft acquired LinkedIn, I interviewed there. I thought I had done well. I had strong technical rounds, clear communication, and relevant experience. But I didn’t get the offer.

A mentor who worked at LinkedIn took 30 minutes to debrief with me. His feedback changed the way I think about interviews.

He said, “Paras, you probably passed the tech and case study rounds. But in the other interviews, you kept talking about healthcare projects. That’s where you were working at the time, but LinkedIn isn’t a healthcare company. You had relevant tech experience too, but it didn’t come through. The stories didn’t land.”

He was right. I had done good work in tech, but I didn’t choose the right stories for that audience. I talked about what I had done recently, not what was relevant to them.

This is the mistake I see over and over again. Candidates tell stories that matter to them, not stories that match what the company is hiring for.

If you want to improve your chances of converting interviews into offers, here’s what you can do:

• Look up what problems the company is solving

• Ask the recruiter what the hiring manager cares most about

• Tailor your talking points to those themes

• Use examples from your past that mirror the challenges they’re facing

Interviewing is hard enough. Don’t make it harder by being generic.

Tell the stories that resonate. Relevance beats recency. Every time.

(Note: written w/ the help of AI tools for polishing/editing language, but the core points and personal story is raw/mine)

First, Remember the Upside

When your inbox floods with urgent requests, remember:

• Trust Indicator – A noisy inbox means leaders believe your insights shift outcomes

• Pivot Power – During market swings, one quick data point can reroute millions in spend

• Spotlight Opportunity – Handled right, ad-hoc work amplifies the team’s strategic value instead of derailing it

The most influential data teams don’t just survive the chaos. They turn it into leverage. Here’s how:

The 3-Step Rhythm for IC’s

1. Acknowledge Fast

• If you’re not heads-down, reply in minutes: “Got it. Routing through intake. Update in 15.”

• Deep-work mode? Auto-reply sets expectations: “In focus block, will triage at 1 p.m.”

• Signal: Responsive, not reactive. There’s a process behind the curtain

2. Fast-Filter (2 minutes or less)

Ask three critical questions:

• Impact: Will this shift ≥ $100k, major risk, or a strategic pivot?

• Urgency: Blocks an exec decision this week?

• Effort: One analyst, half-day or less?

If it’s yes, yes, low → green light. Otherwise log it to the backlog with a clear note on priority and ETA.

3. Execute & Capture

• Ship the smallest artifact that unblocks the request—query, chart, bullet insight

• Log time spent and actual business result

• When a pattern hits its third repeat, productize it (dashboard, alert, or model)

Guard-Rails for Data leaders

To make this sustainable without sacrificing your roadmap:

1. On-call Buffer (10%)

• One rotating team member owns all ad-hoc work for the week

• The rest of your team stays deep-focused on strategic initiatives

• Build this into capacity planning rather than treating it as “extra”

• Success signal: ad-hoc ≤ 10% of total sprint hours

2. Two-Line Responses

Keep communications brief but clear:

• Do Now: “Will ship by EOD; you’ll see a chart in #data channel.

• Backlog: “Queued for sprint starting x”

3. Automate or Educate

• Repeated asks? Either automate or publish a “How to self-serve” Loom/video

• Every quarter, share good vs. bad request examples with PMs and GMs to raise the barBottom Line

4. Equity & Burnout Check

• Track who pulls SWAT duty. Rotate fairly; cap consecutive weeks

• High-visibility ad-hoc shouldn’t always land on the same star

5. Provide Escalation Path

If a requester disagrees with your & team’s filter, escalation is IC → Manager → VP of Data. No hallway lobbying.

Bottom Line

Ad-hoc requests are VIP walk-ins: proof you’re indispensable. Acknowledge fast, filter hard, ship the minimum that matters, automate the repeats, and quantify the impact. That’s how elite data teams turn on-demand chaos into a strategic megaphone without burning the roadmap or the people.

Key Takeaway (60‑Second Version)

• Lead every deck or doc with the impact headline, not the methodology.

• Tell the story with the inverted pyramid: headline → critical metrics → clear recommendation → supporting details.

• Translate tech results into plain‑language business value and one specific ask.

Read these three lines, apply them, and your next presentation will connect.

Where Things Go Sideways and How to Recover

Opening with model architecture

• Why it fails: decision makers tune out

• Fix: begin with “Churn drops eight percent, worth twelve million dollars.”

Hiding the insight on slide 14 or page 5

• Why it fails: skimmers never reach it

• Fix: headline, key metric, single‑sentence recommendation, then evidence.

Sending a dashboard with no story

• Why it fails: readers guess what matters

• Fix: caption every chart with insight, implication, action.

Answering with “it depends”

• Why it fails: feels non‑committal

• Fix: give a directional answer plus the next validation step.

Leaning on jargon

• Why it fails: creates distance

• Fix: swap statistics for plain language.

Writing long emails and docs without an ask

• Why it fails: the request gets lost

• Fix: bullet the message: context, recommendation, decision needed by date

The Inverted Pyramid in Action

1. Headline: one sentence on the business outcome

2. Essential facts: metrics, dollar impact, recommended action

3. Supporting insight: a short explanation of why

4. Details and methods: appendix or final pages for readers who want depth

Use steps 1 and 2 to open any deck or doc.

Pre‑Send Checklist

1. Can someone grasp the point in ten seconds?

2. Is the dollar value or strategic upside clear?

3. Is jargon trimmed to the minimum?

4. Does each visual tell one story?

5. Is the decision request or next step explicit?

Check all five and your analysis is ready.

Here’s something I’ve learned, often the hard way. Enterprise AI is fundamentally different from consumer AI because it’s built on messy, chaotic, and often siloed data. Consumer AI often benefits from data that’s public and widely available, which means it’s been seen, vetted, and scrutinized by a broad community. That doesn’t always make it clean or perfect, but it does give it a level of collective validation. Fair? Feel free to challenge that.

If you’re feeling overwhelmed, you’re not alone. I’m right there with you. This stuff is complex. Here’s how we can begin to make a dent in it:

1. Embrace (and Map) the Chaos

Start simple. Grab a whiteboard or open up a spreadsheet. Map out where your critical data lives. Who owns it? How accessible is it? You might be surprised, and probably frustrated, at what you find. Even just surfacing these patterns can create momentum for change.

I recently did this exercise myself, and it was eye opening. It revealed not just technical gaps but also cultural and organizational barriers. Even partial clarity on your data landscape can go a long way.

2. Build Trust Through Transparency

In consumer AI, trust comes easier because data sources are usually public and already vetted. In enterprises, trust has to be earned. Ever had someone question your insights because they didn’t trust your data? Same here.

One thing that helps: documenting data lineage. Making the process behind your data transparent and understandable gives others more confidence to use it. It’s not flashy work, but it pays off.

3. Unlock Those Legacy Systems (Slowly but Surely)

Legacy systems are the enterprise data leader’s perennial headache. Unlike consumer AI applications, we can’t just plug in modern AI to decades-old systems. Integration is complex, expensive, and culturally challenging.

The good news? You don’t have to fix everything overnight. Focus on integrations that can deliver small but meaningful wins. I’ve found that piloting AI in lower-risk areas can create the traction needed to make larger changes later.

Why This Work Matters

Enterprise AI isn’t about flashy tech. It’s about data clarity, trust, and thoughtful integration. These are not always glamorous problems, but they’re the ones that determine whether AI actually works in practice.

I don’t have all the answers yet and I’m figuring it out as I go. But I believe if we lean into these foundations bit by bit, we’ll move closer to meaningful impact. If you’re in the thick of it too, let’s connect and share what’s working. This is a journey, and we’re all still learning.

1. Turn Vague Requests into Testable Bets

What to do: Translate every “Why is churn up?” or “Can we use AI?” into a hypothesis, a metric, a decision owner, and a time‑box.

Why it matters: Sharp framing kills scope creep and keeps the crew focused on work that changes the business.

Try it today: Rewrite the next ask in Slack, grab a quick thumbs‑up, and only then open the notebook.

2. Treat Analysis Like Shipping Product

What to do: Put notebooks and SQL in version control, demand code review, and add a unit test that fails if row counts swing ten percent.

Why it matters: Reproducible pipelines build instant trust and prevent “it worked on my laptop” disasters.

Try it today: Move one critical notebook into the repo and open a pull request before lunch.

3. Run a 15‑Minute Pre‑Mortem on Signal vs Noise

What to do: Before launch, list the top three threats to validity: sample bias, tracking gaps, seasonality. Mitigate or escalate.

Why it matters: A shaky experiment torpedoes credibility faster than a buggy dashboard.

Try it today: Block the calendar for a quick huddle and refuse to ship until risks are addressed.

4. Automate the “Help Desk” and Buy Back Time

What to do: Standardize repeat questions, automate pipelines, and publish one self‑serve dashboard.

Why it matters: Automation frees brain space for deeper, career‑defining projects.

Try it today: List the five most common pings, answer them in a dashboard, record a three‑minute video walkthrough, and watch ad‑hoc requests drop.

5. Tell the Story in Cash and Risk, Not Accuracy

What to do: Package every outcome as dollars earned or risk avoided. Skip the accuracy flex.

Why it matters: Execs fund clear ROI, not cool algorithms.

Try it today: After each delivery, fire off a three‑bullet update: decision enabled, dollar impact or risk avoided, next step. Collect them in a living “wins” doc for QBRs and promo packets.

Bonus: Put AI on Your Bench, Not on a Pedestal

What to do: Treat generative AI like a junior analyst who is great at first drafts, dangerous without review. Pick one repetitive task (doc summaries, SQL boilerplate, slide headlines) and co‑pilot it with an LLM. Layer in human checks and clear data‑privacy guardrails.

Why it matters: You unlock speed today while training the team to wield AI safely as the tech races ahead. Future‑proof your skill set.

Try it today: Spin up a private chat sandbox, feed it last month’s experiment readout, and have it draft the exec summary. Tweak for accuracy, then time how long it saved you. Roll the playbook to the team.

Bottom line

You’re no longer a super‑IC. You’re the translator, quality gate, and leverage engine. Nail these five plays and your tiny team and your career will scale faster than any model you ship.

The 3 P’s framework—People, Platform, and Process—offers a practical way to build a data-led culture that consistently turns analytics into real business results.

People: The Foundation of a Data-Led Culture

A data-led organization starts with people who are empowered to use data in their decision-making. Executive sponsorship is powerful. When leaders champion data, it sets the tone for everyone else. But being data-led is more than just using dashboards. It’s about encouraging everyone, from executives to frontline teams, to ask better questions, challenge assumptions, and act on insights. If your leadership isn’t fully bought in, start small. Build a coalition of data advocates, highlight early wins, and create a culture where curiosity and evidence are valued.

Platform: Enabling Access and Action

Your platform is the set of tools and infrastructure that puts data into people’s hands. For small teams, this might mean spreadsheets and ad hoc queries. As your organization grows, you’ll need more robust tools that offer self-service analytics, data governance, and integration across departments. The best platform is one that people actually use and trust. It should make it easy to find, understand, and act on data, supporting the processes you’ve put in place.

Process: Turning Insights into Action

Process is where everything comes together. Well-defined processes ensure that analytics projects align with business strategy, data definitions stay consistent, and teams know how to get support. In a data-led company, processes are designed to make data actionable, not just available. This means prioritizing analytics requests based on impact, establishing clear ownership of metrics, and maintaining high data quality. Processes should also encourage experimentation and learning, making it easy to test new ideas and adapt as new insights emerge.

Putting the 3 P’s into Practice

• Identify your biggest opportunity. Assess your organization across People, Platform, and Process. Focus on the area where improvement will have the highest impact.

• Iterate and evolve. Building a data-led culture is a journey. Keep refining your approach as your business and needs change.

• Adapt to your context. The 3 P’s framework works for entire organizations or individual departments. Even if you don’t control every tool, you can always influence people and process.

What’s New for Data-Led Organizations?

• Use data as a guide, not the only voice. Pair analytics with business context and human judgment.

• Empower everyone to challenge assumptions and ask “why,” not just “what.”

• Focus on turning insights into action, not just reporting metrics.

• Encourage experimentation and learning from the data, not just following it blindly.

If you want your organization to make smarter, faster, and more confident decisions, focus on people, build the right platform, and never underestimate the power of great processes. Let data lead you, but don’t let it be the only voice in the room.

If you want to grow faster, build smarter, and avoid painful mistakes, you need to share your work before it’s perfect. Not after it’s launched. Not when it’s too late to pivot. The earlier, the better.

The Common Mistake: Trying to Get It Right Before You Share

Here’s what often happens. You work heads-down. You polish everything. You finally present your work. And then someone asks, “Wait, is this solving the right problem?”

A good example? I once saw someone build a beautifully structured experiment to re-engage lapsed sellers. The logic was solid. Metrics looked great. But it was way too early for that level of rigor. A simple email test could have given us quick signal. Instead, we waited too long to learn something basic. In early-stage work, fast signal matters more than polished certainty.

How to Run a Good Feedback Session Without Making It Awkward

Here’s a simple way to run a quick 30-minute work-in-progress (WIP) review that actually helps:

1. Share unfinished work. Show your draft, 60% complete is totally fine — and don’t wait for final product. Feedback is most useful when the work is still flexible.

2. Ask specific questions. Instead of saying, “What do you think?”, try asking:

• “Does my success metric match the business goal?”

• “Where would you challenge this approach?”

• “What am I assuming here that might not hold?”

3. Focus on thinking, not formatting. Don’t worry about typos or perfect charts. Ask teammates to sanity check your logic, not your grammar.

4. Write down what you hear and follow up. When someone gives you feedback, take notes. Make updates. Let them know what you changed. It builds trust.

What If My Team Doesn’t Do This?

You don’t need a formal process to get feedback. Ask two or three teammates if they’d be open to trading feedback every other week. Take turns walking through what you’re working on. Even quick, casual sessions can make a huge difference.

Why This Works

The best data scientists aren’t perfect. They just learn faster. And they get better because they invite critique instead of avoiding it.

So next time you’re tempted to perfect your work before showing it to anyone, try this instead: share early, ask good questions, and treat feedback as a career accelerant.

Because great work isn’t built in isolation

Below are the most common reasons WBRs flop—and how to turn them into a true driver of business results.

Why WBRs Fail

1. Lack of Clarity Teams don’t have well-defined OKRs (Objectives and Key Results), leading to confusion about what success looks like.

2. No Metric Ownership When everyone “owns” a metric, no one really does.

3. Excessive Pre-Prep A single team (often Analytics or Data Science) scrambles to pull data—unsustainable over time.

4. Reviewing Every Metric Meetings get bogged down in endless details instead of focusing on key variances or anomalies.

5. Poor Context Attendees don’t prep beforehand and end up skimming through dashboards live (wasting everyone’s time).

6. Wrong People in the Room Critical metric owners are missing, stalling key decisions.

How to Fix It: Your Action Plan

1. Clarify OKRs Align each team’s goals with the company’s objectives so everyone knows what really matters.

2. Assign Single-Threaded Owners Give each metric one clearly identified owner—no ambiguity, no confusion.

3. Zero In on Variances Skip the metrics that are performing as expected. Focus your meeting on what’s off-track or unexpectedly spiking.

4. Prep Asynchronously Share dashboards or notes 24 hours in advance, so the live discussion is short, sweet, and to the point.

5. Invite the Right Folks Make sure the people who can explain metric variances or make decisions are actually at the table.

6. Standardize Data Sharing Everyone should have easy, unblocked access to the data needed.

7. Track Follow-Ups Each discussion point should have an owner and a due date to ensure accountability.

8. Encourage Open Dialogue A culture of candor and curiosity drives deeper insights—and better decisions.

A Streamlined WBR in Practice

• 15 Minutes (Async): Attendees review metrics, anomalies, and context in a shared dashboard.

• 30 Minutes (Live): Discuss only the most critical variances—why they happened and what’s next.

This structure lightens the prep load and keeps the live session laser-focused on solutions, not just data regurgitation.

Final Thoughts (and a Personal Note)

As a data leader who spent nearly 5 years at Amazon, I’ve seen firsthand how a well-run WBR can transform an organization. Since leaving Amazon, I’ve helped multiple teams turn around broken WBR processes—driving cultural change, instilling real accountability, and freeing teams from time-draining meeting overload. That journey is the inspiration for this post.

The WBR isn’t a magic wand. It’s part of a broader operating rhythm—annual planning, monthly reviews, innovation mechanisms, etc. But if you nail these core principles (clarity, ownership, variance focus, accountability), you’ll transform your WBR from a dreaded meeting into a powerful driver of action.

Resource

My favorite resource on this topic is written by former Amazon Executives Colin Bryar & Bill Carr . If you’re looking for deeper insights on how Amazon structures its WBR, check out their user manual: Working Backwards WBR User Manual

Act like you already have the role you aspire to—demonstrate the mindset, behaviors, and leadership you’d show if that promotion (or dream job) was already yours.

This mentality shift not only changes how others perceive you, but also how you approach opportunities and solve problems. Below are some modern career growth “hacks” that can help you level up—especially relevant in #DataAnalytics and #DataScience roles.

1. Act “As If” You’re Already There

• Mentally Upgrade Your Role: Ask yourself, “If I were the Team Lead or Senior Data Scientist, how would I approach this project, meeting, or cross-functional collaboration?” Then take action accordingly (e.g., propose solutions, share data‐driven insights, mentor junior team members).

• Proactive Problem‐Solving: Don’t wait for someone to assign tasks. Identify gaps or challenges in your department and pitch data‐backed solutions.

Why It Works: It signals you’re ready for greater responsibility and leadership—even before an official title change.

2. Share Your Expertise Openly

• Micro Thought Leadership: On your internal Slack, project management platforms, or LinkedIn, share quick insights or tips you’ve learned from your day‐to‐day.

• Knowledge Sessions: Offer to host short workshops or “lunch & learn” sessions on cool analytics topics, new Python libraries, or data visualization tricks.

• Contribute to GitHub: If you can open‐source parts of your code (within company policy), you’ll build credibility in the broader data community.

Why It Works: Demonstrating expertise (and generosity in teaching others) positions you as a go-to resource, increasing visibility within and outside your organization.

3. Build a Future‐Focused Skill Stack

• Identify In‐Demand Technologies: Stay ahead on AI/ML frameworks, data engineering stacks, or trending analytics tools (e.g., Spark, dbt, Snowflake, etc.).

• Micro-Certifications: Platforms like Coursera, Udemy, or LinkedIn Learning offer bite‐size specialized courses. Build a portfolio that reflects breadth + depth.

• Personal Projects: Work on side projects that showcase your ability to solve real‐world problems—like analyzing public datasets or building predictive models for interesting challenges.

Why It Works: Continual upskilling ensures you remain relevant in a rapidly changing data landscape. Hiring managers and leaders notice proactive learners.

4. Cultivate a Mentor & Peer Network

• Internal Mentors: Seek out someone in a higher data/analytics role who’s walked a similar path. Ask for advice on specific challenges, next‐level leadership skills, or domain knowledge.

• External Communities: Join data science meetups, Slack channels, or LinkedIn groups (#DataScience, #Analytics). Engaging in Q&A threads or events expands your horizons and your professional circle.

• Peer Coaching: Align with a colleague at a similar level—swap feedback on projects or share learnings from conferences and workshops.

Why It Works: Access to a broader network helps you learn faster, get introduced to new opportunities, and gain perspective beyond your immediate team.

5. Communicate Like a Leader

• Data Storytelling: It’s not enough to crunch numbers—learn to frame insights in a compelling narrative. Tailor your language for non‐technical stakeholders (e.g., business managers, executives).

• Take (or Make) the Stage: Offer to present in team stand‐ups or all‐hands meetings. Volunteer to represent your team’s results in cross‐dept forums.

• Clear & Concise: Sharpen your writing. Executive summaries or dashboards should be easily digestible so busy stakeholders can make swift decisions.

Why It Works: Communication is a massive differentiator in data roles. The easier you make it for others to understand the “why” and the “so what,” the more impactful you become.

6. Seek “Stretch” Assignments

• Cross‐Functional Projects: Volunteer for tasks that require interdepartmental collaboration. This broadens your organizational awareness and grows your influence.

• Take the Initiative: Notice a bottleneck in data pipelines? Step up to propose or pilot a solution.

• Collaborate with Senior Stakeholders: If you get the chance to partner with directors or VPs on a data initiative, seize it. You’ll learn from their perspective and showcase your skills at higher levels.

Why It Works: Stretch projects accelerate your experience and demonstrate readiness for more complex roles—both technical and leadership.

7. Reflect & Celebrate Wins

• Track Your Achievements: Keep a simple “brag document” or personal KPI log. Jot down each project’s outcomes, metrics improved, or cost savings.

• Visibility in 1:1s: During check‐ins with your manager, highlight not just what you did, but why it mattered to the business.

• Give Yourself Credit: This confidence flows into how you show up, making it easier to own that “next‐level” mindset.

Why It Works: Having a record of your achievements helps articulate your impact, which is invaluable for performance reviews, promotions, or job interviews.

Conclusion

Modern career acceleration is less about wearing a nice suit and more about fully inhabiting the skills, mindset, and leadership behaviors associated with your next career step—even if your current title hasn’t caught up yet. By staying curious, sharing knowledge, and proactively shaping your role, you make it hard for decision‐makers to ignore your growth and impact.

Start acting as if you’ve already landed that next position. Before you know it, your manager, peers, and the wider network will see you in that light too—and official opportunities often follow.

“Dress for the job you want” → Behave, create impact, and communicate like you already have it.

Good luck on the journey—here’s to your continued growth in #DataAnalytics and #DataScience!

Originally published by Paras Doshi on LinkedIn here

This newsletter aims to promote continuous learning for data science and engineering professionals. To achieve this goal, I’ll share articles from various sources I found interesting. The following 5 articles/videos made the cut for today’s newsletter.

1. Data Contracts 101 by Aurimas Griciūnas

The concept of a Data Contract is an agreement between Data Producers and Data Consumers on the schema, SLAs, semantics, lineage, and other details of the data being produced. Data Contracts should be enforced to ensure data quality, prevent unexpected outages, enforce ownership of data, improve scalability, and reduce the intermediate data handover layer. An example implementation of Data Contract Enforcement involves schema changes in a git repository, data validation against schemas in the Schema Registry, pushing validated data to a Validated Data Topic, validating data against additional SLAs, and alerting Producers and Consumers to any SLA breaches. Read more here

2. A brief history of Data at Coinbase by Michael Li

The article provides a brief history of data and its importance in the development of Coinbase, a cryptocurrency exchange platform. The author explains how the concept of data has evolved over time and how Coinbase has utilized data-driven decision-making to improve its platform and expand its user base. The article also discusses the potential of Web3, a new decentralized web infrastructure, and how it can revolutionize the way data is stored, shared, and used. The author concludes by emphasizing the importance of data in the growth of Coinbase and the potential of Web3 to transform the future of data.
Read more here

3. How to use the Snowflake Query Profile by Ian Whitestone

The article explains how to use the Snowflake Query Profile, a feature of the Snowflake cloud data platform, to diagnose and optimize SQL queries. The Query Profile provides detailed information about the query execution plan, including the amount of time spent on each operation, the number of rows processed, and the resources consumed. The article walks through the steps of running a query and analyzing the Query Profile to identify potential bottlenecks or areas for improvement. The author also provides tips for optimizing queries based on the information provided by the Query Profile. Overall, the article offers a useful guide for developers and data analysts looking to improve the performance of their Snowflake SQL queries. Read more here

4. Building Modern Data Teams by Pedram Navid:

The article discusses the characteristics of modern data teams and the key roles involved in building and managing data infrastructure. The author argues that data teams should be cross-functional, collaborative, and focused on delivering business value through data insights. The article identifies several key roles in modern data teams, including data engineers, data analysts, data scientists, and product managers. The author provides an overview of the responsibilities and skills required for each role and emphasizes the importance of communication and collaboration between team members. The article also highlights some of the challenges faced by data teams, such as data quality and security, and provides tips for overcoming these challenges. Overall, the article provides a useful perspective on the structure and function of modern data teams. Read more here

5. How to Prioritize Analytical Work by Elvis Dieguez

The article provides tips on how to prioritize analytical work effectively. The author suggests that prioritization should be based on the business impact of the analytical work, as well as it’s level of complexity and the urgency of the request. The article recommends creating a prioritization matrix that takes these factors into account and prioritizing work based on its placement in the matrix. The author also emphasizes the importance of communication and collaboration with stakeholders to ensure that priorities are aligned with business needs. Additionally, the article provides some tips for managing a backlog of analytical work and for tracking progress and results. Overall, the article offers practical advice for data analysts and other professionals responsible for managing analytical workloads. Read more here

Thanks for reading! Now it’s your turn: Which article did you love the most and why?

1. People management activities: This bucket includes tasks where you work backwards from keeping and growing the folks on your team. E.g. weekly 1:1’s, Team meetings, Career growth sync’s, Promotion process, etc.

2. Hiring & team building activities: This bucket includes tasks related to hiring new folks or backfilling existing roles on your team. This also includes tasks to continue to have a culture & structure on your team that other folks would love to be part of.

3. Partnerships and stakeholder management: This bucket include tasks to build partnerships and trust with teams that your team’s success depends on. This also includes proactively managing relationships and obsessing over the needs of our stakeholders by meeting them, being in org-level forums, etc.

4. Building & driving tech vision: This bucket includes a) forming a tech vision for your team that you can work backwards from b) putting mechanisms to drive towards that vision and empowering the team to be effective and efficient. This could take various forms but roadmap, annual planning, head-count planning, prioritization mechanisms, tech architecture reviews, etc. are part of this.

5. Strategic Initiatives: I always have some or the other strategic initiative going on. These are initiatives that will have a step-change improvement for company, org and the team.

These buckets can be further broken down into people (people management, hiring), process (partnerships, stakeholder management) and platform (tech vision, strategic initiatives) and maps very well to mental model to building data driven companies which I have written about before here: https://insightextractor.com/2015/12/29/building-data-driven-companies-3-ps-framework/

Hope that helps and I would love your feedback, comments below!

(1) How to drive a team vision as First-time manager?

Paras recalls how he successfully drove a team vision as a first-time manager who too over a team without the vision and roadmap. Read here: https://www.platohq.com/resources/how-to-drive-a-team-vision-as-a-first-time-manager-1086187395

(2) How to unlock the potential of your average engineer?

Paras discusses how to unlock the pontential of an average performing engineer and encourage them to be more proactive and autonomous. Read here: https://www.platohq.com/resources/how-to-unlock-the-potential-of-your-average-engineer-658984829

(3) Managing career growh of your reports?

Paras tells how he approaches career growth of his reports by dedicating time to exclusively talk about their careers. Read here: https://www.platohq.com/resources/managing-career-growth-of-your-reports-1306933855