Betty Cherif
CUSTOM JOB APPLICATIONS
Just a girl who believes great companies deserve great applications
👋 Hi Chewy,
I didn’t just apply, I built this.
So you can hire me on proof, not promise.
Business Intelligence Engineer @Chewy
The Context
I know how much pressure there is for business leaders to hire the right person, especially when the mission matters and the workload is real.You want someone who’s not just capable, but dangerous in the best way—and deeply aligned with your mission.
You also want proof, not promise.
I built this website for you. To make it easy to evaluate my candidacy.I reverse-engineered your job description to build a working tool-something that directly answers what you're hiring for.
I don't have access to your SQL or internal data so I got creative with public data. Figuring I had about 10 days before the posting closed, I drank tons of coffee and wrote 900+ lines of code to build a fully functional Business Intelligence tool for Chewy.Just to make it easy for you to say: “Yep. She thinks like us. And she’s already in motion.”
THE TOOL
A Real Tool That Listens to Pet Parents on Reddit
You already know Reddit is where people go to be honest.No brand filters. No sponsored content. Just raw, unfiltered feedback from pet parents — sharing what they really think about companies like Chewy, Petco, and PetSmart.That’s exactly why I started there.This tool reads Reddit to tell you exactly how pet parents feel about you and your competition. It collects real-time feedback from pet owners, identifies which brand they’re talking about, and uses sentiment models to summarize their strongest opinions.So if you’re wondering:
• What are customers praising Chewy for this month?
• Why are people switching to Petco?
• How strongly do they feel?This tool has the answers. Any day you need them. Without waiting for a quarterly survey
What You’ll See When You Use It
1. Finds the posts where people rave about Chewy — or raise concerns
In real time, the tool pulls Reddit posts and comments mentioning Chewy and filters them with GPT. It separates emotional reactions from noise and organizes them by brand, so you only see what matters.2. Compares Chewy to Petco and PetSmart head-to-head
See exactly how Chewy stacks up in terms of sentiment, post volume, and clarity of praise or complaints — month by month.3. Visualizes trends and top moments
Track sentiment over time and identify the highest-praise and harshest-criticism months. One click reveals a GPT-written summary of what people loved or hated that month.4. Shows how customers talk when they actually care
This tool only includes posts with strong, clear emotional language — not generic mentions. You get signal, not noise.
Try It Yourself
🛠️ MVP built in 10 days. Scaling it would be a pleasure.
🖥️ Note: For the smoothest experience, try the tool on desktop.
What This Means For You
What This Means for You
TOOLS USED
• Python – core language
• Streamlit – app framework
• PRAW – Reddit API wrapper
• spaCy – text preprocessing
• RoBERTa – sentiment scoring
• OpenAI GPT-3.5 – brand relevance checks + topic labeling
• Plotly & Matplotlib – visuals
• ETL Pipeline – Reddit data is extracted, cleaned, transformed, and loaded into a Streamlit interface for interactive analysis
All analysis runs client-side. GPT is used only for post-labeling, not for classification or logic.
DATA CLEANING
I spent dozens of hours just cleaning the Reddit data — because sentiment models are only useful if you know the posts are actually about the brand.
That meant:
• Pulling from 12+ keyword variants (not just “chewy”) to catch real brand discussions — not “this baguette is chewy” or “this candy is too chewy”
• Filtering out irrelevant matches (GameStop, Ryan Cohen)
• Removing 'Amazon' and 'Stella and Chewy's' mentions that skewed the signal
• Using GPT to confirm post-level brand relevance (not just keyword presence)
WHAT’S NEXT
In a future version (Zoom in on Chewy), I plan to cluster Chewy mentions using BERTopic and tag them by feature (e.g. Autoship, Pharmacy, Portraits) — giving the team more granular insight into what’s driving brand sentiment.
NOTE: Minor differences in analysis may appear across runs because Reddit search results are dynamic and can vary over time.
If you're a recruiter, this project aligns closely with the responsibilities listed in the job description for BI Engineer II, Merchandise Analytics.
I submitted my formal application on May 21, including a resume and cover letter — this project is just a bonus, to show what I can do.
I created this tool without access to Chewy's internal systems or SQL databases, but I’m fluent in SQL and use it regularly. Instead, I pulled public data and built an ETL pipeline using external APIs and Python libraries like pandas, spaCy, Plotly, and OpenAI to simulate the kind of customer sentiment and trend analysis I’d love to do at Chewy.
I believe it reflects the same core skills the role emphasizes: working with messy data, building clear visualizations, and using external insight to drive better decisions.
I know Petco and PetSmart are solid competitors — but Chewy vs. Amazon is the big question.
Reddit talks about Amazon a lot, but since Amazon spans thousands of product categories, it’s nearly impossible to isolate posts that are truly relevant to pet care. I tested a few variations — like searching “Amazon dog” or “Amazon cat” — but even those returned everything from chew toys to patio furniture. That makes meaningful comparison noisy and inconclusive.
But I'm not giving up.
I'm currently exploring a few approaches:
• Identifying Reddit users who mention both Chewy and Amazon in the same post
• Using product-related keywords (like “dog bed Amazon”) to find higher-intent content
The goal: help Chewy leadership distinguish between volume and value — so brand comparisons aren’t made on noise but signal.
NOTE: Minor differences in analysis may appear across runs because Reddit search results are dynamic and can vary over time.
P.S. I’m covering the cost of both hosting and backend AI services out of pocket (Render + OpenAI) because when I say I'm all in, I mean it
Meet the Builder
Hi, I’m Betty — a BI Engineer who spent the last year going deep with Python. Before that, I built pricing tools at Liberty Mutual, where I realized I loved engineering more than standalone data analysis.This project is part of my self-funded career break — a hands-on, high-intensity stretch that I approached like a graduate-level sprint in engineering. During that time, I learned Streamlit, Dash, Plotly, and GPT APIs. I was already a systems builder, known for creating pricing tools and automating workflows. But I wanted to deepen my technical fluency — to build with greater speed, clarity, and code.In insurance, my work centered on the complexity of actuarial forecasting and pricing. While there was some engineering involved, the emphasis was often on the underlying calculations — not on the internal structure or architecture of the tools. I knew I wanted to shift that balance: to focus more fully on engineering, and to build tools that could serve real business users — especially in high-velocity, customer-facing sectors like retail.When I saw the role at Chewy, it immediately stood out — a strong match for my skills and interests at a company I genuinely admire. So instead of just applying, I built something: a custom tool that answers brand questions using live public data.I’d love to bring this mix of precision, creativity, and engineering to your team.
The Problems I Can Solve For You
Yes, I can build interactive dashboards and reports but that’s not where my value ends.What I really do is build living tools that sit on top of your models — tools that help business leaders understand what their data science teams are really saying, in real time.Because I come from a technical background, I know how to sit right next to a data scientist and translate their work into a usable interface — something that turns models into decision-ready, business-friendly applications.That means:
• Your VPs don’t have to ask for “just one more view”
• Your directors don’t need a meeting to interpret the math
• Your team cuts out the game of telephone between science and strategyWhether it’s churn prediction, merchandising forecasts, or autoship optimization, I build tools that make the insight click — instantly, and in the language of the business.
Want to Hire Me?
I built this because I believe great, mission-driven companies deserve this level of effort. If you’re in a position to move strong people into the room, this is for you.Note: I applied for the Business Intelligence Engineer - Merchandise Analytics (R26417) role
A Quick Note on Alignment
I applied to the only matching BI role open right now
even though it’s titled below the level I operate at.I didn’t build this to prove I’m a senior BI engineer.
I built it because I already am one.If you value clarity, precision, and initiative—and rewarding at the appropriate level—
I’ll return that by being RELENTLESS in building the tools that help you move faster, think clearly, and hit your profit goals.
June 16th
I’ve started working on my next custom application,
but I prefer to focus on one company at a time.This site is dedicated to Chewy through June 16
What's Next for Me?
I'm all about finding the right role at the right company. And my list of favorite companies is long.🐾 But it sure would be cool to land the one that makes pets happy. 🐾
... not even a dog mom yet, just an over-involved dog auntie