Upfynd is looking for a Machine Learning Engineer to keep our rankings honest and useful. On Upfynd, products rise because real people upvote and review them, so the signals have to be clean. You will build the models that catch manipulation and help the best new software get noticed.
Why this role matters
A ranking that can be gamed is worthless. Your models will separate genuine community interest from vote rings, bulk accounts and paid reviews, and they will help decide which new launches deserve a place on the front page.
What your week looks like
- Build models that detect suspicious voting patterns, bulk accounts and coordinated activity.
- Detect spam, fake and incentivised reviews before they affect rankings.
- Improve trending and launch rankings using votes, reviews and engagement over time.
- Engineer features from user, product and activity data and keep training data well documented.
- Deploy models to production, monitor drift and retrain when patterns change.
- Explain model decisions clearly so moderators can act on them with confidence.
You will do well here if you have
- 3-5 years building and deploying machine learning models in production.
- Strong Python and experience with scikit-learn, XGBoost or PyTorch.
- Experience with anomaly detection, fraud detection, ranking or recommendations.
- Good SQL and experience with feature pipelines and model monitoring.
- Clear thinking about fairness, false positives and explainability.
Nice to have
- Experience with graph-based methods for detecting coordinated behaviour.
- Experience in trust and safety or marketplace integrity.
- Familiarity with MLOps tools such as MLflow.
Your first 30, 90 and 180 days
- 30 days: You understand Upfynd's ranking signals and how manipulation attempts usually look.
- 90 days: You have shipped a first integrity model with moderators using its output.
- 180 days: You own vote and review integrity, with clearly less manipulation reaching the rankings.
Where and how you will work
This is a full-time, onsite role at our office on the 11th floor of O-Hub, Infocity, Bhubaneswar. You will sit with the people you build with, which keeps decisions quick and conversations easy. Your everyday tools will include GitHub, Python, scikit-learn, PyTorch and Docker.
About Upfynd
Upfynd is a directory of new software, ranked by the people who actually use it. Every person gets one upvote per product, reviews come from real users, a person on our team checks every listing before it goes live, and nobody can pay to move up. Makers list and claim their products for free, and anything sponsored is clearly labelled and never touches rankings.
How to apply
Send us your CV along with a short note on a model you deployed and how you measured its impact. Just like every listing on Upfynd, every application is read by a real person, and we reply to everyone who applies.
Skills and tools
- Python
- scikit-learn
- PyTorch
- XGBoost
- Anomaly Detection
- Fraud Detection
- Learning to Rank
- SQL
- MLOps