全职 Senior Data Scientist 工作, 薪水, PathIQ 公司招聘中 - Ricebowl

Data Scientist — TrustIQ

Company: PathIQ Pte. Ltd.

Location: Singapore (on-site / hybrid)

Type: Full-time, permanent

Reports to: Head of TrustIQ

ABOUT PATHIQ

PathIQ is a Singapore-headquartered data and AI company. TrustIQ is our identity and trust API: a client sends a hashed email, hashed phone number or mobile ad ID, and we return consented, privacy-compliant signals that help banks, lenders and credit platforms decide who they are dealing with.

We are small, fast, profitable and building the trust layer that financial services in emerging and digital-first markets have been missing.

THE ROLE

You turn raw signals into scores that financial institutions will rely on.

TrustIQ is moving from returning attributes to returning scores. Working on top of our AWS data platform and ID graph, you will design, build and validate the models behind those scores, and prove to clients that they hold up. This is applied, commercial data science — the output is a score that is accurate, explainable and cheap to serve, not a paper.

WHAT YOU'LL OWN

Score design & modelling

• Design the trust and risk scores TrustIQ returns — define what each score means, what feeds it and how it is calibrated

• Build and tune models with per-attribute, per-country weights, using regression, gradient boosting and back-testing

• Evaluate modelling approaches (classic ML, LLM-based, hybrid) on accuracy, cost and explainability, and recommend what we ship

Feature engineering

• Turn mixed raw sources — programmatic, SDK, telco, transaction and commerce data — into clean, stable features

• Develop signals from device, location, behavioural, spend and identity-stability patterns

• Work with data engineering on the daily batch that stages, cleans and scores data into a precomputed feature and score table

Validation, bias & explainability

• Build the statistical framework that measures score accuracy, stability and confidence by market and segment

• Back-test scores against client outcomes and third-party benchmarks

• Detect and quantify bias — device skew, coverage gaps, demographic and geographic bias — and document it

• Produce model documentation that stands up to a regulated client's model-risk and compliance review

Partner data evaluation

• Assess prospective data sources before we commit — coverage, uniqueness, overlap, accuracy, refresh rate

• Quantify what each source adds to match rate and score quality, and what it duplicates

• Monitor ingested data for drift, degradation and anomalies over time

Accuracy & cost per lookup

• Own the trade-off between score quality and cost per API call — decide which inputs earn their place

• Track match rate, coverage and score performance in production and feed findings back into the models

Client & commercial support

• Design and analyse proof-of-concept tests with financial services clients, and turn results into evidence they can act on

• Present methodology credibly to risk, fraud and data teams

• Support product with feasibility, packaging and pricing input

WHAT YOU'LL NEED

Essential

• 4+ years applied data science, ideally in credit risk, fraud, identity, fintech, telco or alternative data

• Strong Python (pandas, scikit-learn, numpy) and strong SQL

• Hands-on experience building and validating scoring or classification models — calibration, back-testing, AUC/KS/Gini, stability monitoring

• Solid grounding in statistics: sampling, significance testing, experimental design, bias correction

• Able to work with large, messy datasets in a cloud environment (AWS preferred) without needing everything pre-prepared for you

• Can explain a model to a non-technical or compliance audience and have them trust it

• Eligible to work in Singapore, or eligible for an Employment Pass

Advantageous

• Experience with credit scoring, fraud detection or KYC / onboarding risk models

• Identity resolution or ID graph experience (hashed email, phone, device IDs)

• Familiarity with model governance and explainability in regulated environments, and with data protection regimes such as PDPA or GDPR

• PySpark for large-scale feature work

• Experience applying LLMs or foundation models to structured or tabular data

• Experience working with banks, lenders or payments companies in APAC

FIRST 6 MONTHS — WHAT SUCCESS LOOKS LIKE

• 30 days: Deep familiarity with our data assets, ID graph and live client proof-of-concepts; initial assessment of source quality, coverage and gaps

• 90 days: Validation framework in place; first scores built and back-tested for priority markets; modelling approach recommended

• 180 days: Scores live in the TrustIQ feed with measured accuracy, client-validated in at least one proof-of-concept, and a lower cost per lookup

WHY PATHIQ

• Build the scoring layer of a commercially live product from the start, with direct access to the Head of TrustIQ and the founders

• A profitable, growing business — not a burn-and-hope startup

• Rich, large-scale identity, behavioural and location data across APAC and global markets

• Competitive package and profit share

PathIQ is an equal opportunity employer.