Breaking into data science feels harder than ever. Thousands of skilled candidates compete for a few openings, with job descriptions that seem to require Ph.D.-level experience for entry-level roles. Even if you have the right skills, getting past the resume screen or landing an interview often feels like pure luck. In a market this tough, how do you stand out?
While landing a callback does have an element of being in the right place at the right time, there are still three key factors you can control to significantly improve your chances:
- Selecting relevant projects
- Creating a strong resume and LinkedIn profile
- Building a network that can refer you
Let’s dive into each of these.
Choosing a Relevant Project
The key capabilities you want to demonstrate via your project for Data Science roles are:
- Ability to extract modeling insights and engineer features – this includes cleaning data (outliers, missing values, imbalance, encoding), understanding data nuances (skewed distribution, dependencies), and engineering predictive features
- Expertise with fitting and fine-tuning models – this means converting business problems to ML problems, choosing model metrics, choosing models, and fine-tuning a model
- Ability to analyze model errors, and improve on v1 models – this means understanding where and why the model is falling short, and deciding between investing in techniques/models to improve performance OR revisiting data to improve performance
- Familiarity of designing production model train/inference pipelines – ability to design (typically you will have engineering help/guidance to write production-quality code) production ML pipeline (i.e., Cloud Stack), AWS/GCP/Azure (focus on compute, storage, and serving) Orchestration Layer (Airflow, SageMaker Pipelines, etc.) and Containerization (Docker)
Kaggle is a great place to find projects. Choose projects that were part of competitions that had large prize money (as it related to the importance of business problems) in the last five years. A few example are:
Payment Fraud Detection
FinTech is one of the few fields that consistently hires data-folk, and fraud detection remains one of the most common use cases. The dataset is real-world e-commerce data, and the discussion board is littered with directions on feature engineering that you can tap into!
Walmart Sales Forecasting
This dataset provides ample opportunity to showcase ability to clean data (outliers), fit and tune models (can experiment with statistical models like ARIMA to DNN models like LSTM), and improve on v1 models by adding external data (sales info, SNAP days, weather, etc.).
This is also a good project to build out a batch model prediction pipeline for, and host results on, a Tableau dashboard — the insights from which could help a merchandiser decide their upcoming assortment, or help marketing decide what deals to push.
Quora Insincere Question Identification
A project to showcase your NLP knowledge, such as: cleaning text, handling embeddings and extracting semantic meaning. Unlike typical NLP projects, this project provides ample room to analyze errors, dive deep into peculiarities of the English language, make hypotheses on how to account for these peculiarities and improve a v1 model. This makes for great conversation during interviews!
H&M Fashion Recommendations
Great project to stand out in the RecSys space. This dataset allows you to start with basic methods — content/collaborative filtering — and slowly advance to state-of-the art methods, such as two-tower networks. Additionally, this dataset has images, allowing you to demonstrate ability to handle multi-modal data.
This is also a good project to create an inference pipeline (i.e., train model on data you have, a customer with specific customer ID hits the model API endpoint and you serve the customer a “landing page” featuring a set of items personalized to customer). You can even build out multiple carousel like:
- Customers who bought this also bought (”Cross Selling”)
- Styles you might like (based on their preferences)
Crafting your Resume
Once you have an amazing portfolio, the next step is to market it well. This starts by writing a stellar resume. The three main things when it comes to your resume are structure, content and tailoring.
How to structure
The key sections in your resume are:
● Career Summary
● Experience/Projects
● Education
● Skills
Keep skills towards the end of the resume, since it’s more for an applicant tracking system (ATS) than a human reviewer.
What to write?
Career Summary
A brief paragraph that has, at most, three lines that cover:
- Who you are (e.g., Data Analyst with 3+ years of experience driving decision-making in the supply chain and retail domain).
- Your core skillset (e.g., Skilled in Python, SQL, Tableau, Statistical Analysis, and Cross-collaboration across teams and job functions).
- A career highlight (e.g., Built a hierarchical clustering model (using engagement/demographic data) to segment customers of XYZ, identify high-value customer groups, improve targeted email marketing, increasing CTR by 9% and GMV by 12% (~$15M)).
Experience/Project
The two main things when it comes to writing bullets in your resume are:
- Each bullet should stand on its own, which means the bullet should not depend on the user reading an earlier bullet.
- Each bullet should answer three questions (also called PSI format – problem, solution, impact)
- What business/customer problem was solved?
- What were the solution’s technical details? Be specific with model names (e.g., XGBoost, VGG, Llama) or techniques (e.g., segmentation analysis, RCA analysis, etc.)
- What was the solution’s quantitative impact? If business facing use business metrics, if an academic/personal project compares your solution with a baseline to showcase the quality of your solution.
A few possible follow-up questions you may have are:
How long should a bullet be?
Not more than two lines of a resume. People lose their train of thought when a sentence runs beyond two lines, and hence your value won’t come across.
If I add a problem, solution and impact, won’t the bullet become too long?
No, you should be able to reduce it to under two lines. If not, you may need to break your bullet into multiple bullets or sub-bullets. For instance, when describing how you built a model, the initial bullet can be about the model fitting and tuning, second bullet can be about data cleaning/feature engineering (use ablation study to show value), and last bullet can be about productionalization (use engineering metrics to show value).
What if I don’t know what the impact on business was?
Conservative estimates based on verbal feedback can be used instead.
Skills
In this section, it’s good to have a ‘Competencies’ subsection, where you add keywords relevant to your role (e.g., Python, SQL, Segmentation Analysis, A/B testing, etc.).
How to tailor resumes?
Typically, I don’t advise rewriting all the bullets in your resume for each role, it’s too time consuming. In the current job market, you want to target quite a few applications to improve your chances of a callback. So what should you do instead?
Create Themed Resumes!
Create multiple versions of your resume based on business domains you have worked on. For instance, if my experience is spread across marketing and supply chain, I would create three versions — one focused on marketing, one on supply chain, and one a mix of the two.
Tailoring your Resume
When applying, go through the job description and identify the Keywords in the Job Description (ChatGPT Prompt: Identify all keywords in the following job description that would be relevant for <role name> at <company name>). Then add all applicable keywords to the “competencies” subsection in the “skills” section of your resume. This should help clear any ATS requirements.
After this, identify the key requirements in the job description (usually the top three requirements) and adjust your career summary to suit it as much as possible.
Note: Now there is an exception when you want to tailor your resume. If you have a contact at the company who will reach out to the Hiring Manager on your behalf, then do take a moment to personalize your resume as you see fit, and based on the feedback of the person who will reach out.
Optimizing your LinkedIn Profile
Having a good LinkedIn profile helps improve the chances of getting a callback. So here is a checklist for optimizing your profile:
- A good photo where your face is clearly visible (no AI-generated images)
- A good cover page related to the role you are targeting or your current college/company
- ‘Open to Work’ – visible to only recruiters
- A good tagline (e.g., Business Analyst | Leveraging Data for Actionable Insight | 3yrs+ in Retail, Supply Chain | Masters in Business Analytics | ex-Company)
- A well-written ‘About’ section
- 1st Paragraph – same as the career summary for the resume
- 2nd/3rd Paragraph – PSI format of two of your best projects.
- 4th Paragraph – a motivational line showing your passion for the field (e.g., Here to create an impact! I want to influence the next wave of machine learning products in e-commerce personalization).
- Fill in your experience and projects just like you did on your resume. Sometimes you can’t reveal the business impact you had, in which case you can skip the metrics on LinkedIn.
- Add Keywords to your skills section — this is used by recruiters at times to find candidates. So add the most relevant ones using this: https://linkedin.github.io/future-of-skills/
- If possible, get recommendations (from peers, seniors or professors)
Networking
Connecting on LinkedIn
You have 100 connection requests you can send per week. Make sure to max them out. The connection message can be simple:
Hi, Would like to request a referral for Job ID: ABC. Given my work in <mention a PSI format project that related to the JD>
- LinkedIn connection requests: Out of 100, expect at least 4–6 responses if your profile is strong and messages are personalized.
- If someone accepts your connection, send a polite follow-up message to establish rapport.
Cold Emails
This is also a number-based strategy. For a good format, check out Leon Jose’s post about the same.
In-Person Networking
Attend Conferences, Meetups and Hackathons
- Use platforms like Meetup, Eventbrite, and LinkedIn Events to find in-person or virtual events.
- Major cities often have startup events and industry-specific conferences. Attend to meet recruiters and like-minded professionals.
- Competitions and hackathons are also excellent for showcasing your skills and building connections.
Check out Angelica Spratley’s post for more on finding communities and in-person events.
The Data Science job market is quite competitive in 2025. With low callback rates and interviews tending to drag on for months, as a candidate you may feel quite hopeless at times. However, know that getting hired is a numbers game. As long as you keep applying, keep improving and keep interviewing — eventually you will land a role.
