Anomaly Detection
Designing detection logic for fraud, unusual customer behavior, repayment stress, and operational drift using statistical baselines, ML models, and monitoring feedback.
Financial Data Scientist & Data Science Manager, Nepal
Leading teams that turn financial data into risk intelligence and trustworthy decision systems.
5+ years leading client-facing data implementations in fraud, AML, and credit risk for banking and fintech clients across Nepal and international markets. I lead a 19-person R&D and Fraud Analytics org built on Python, PySpark, SQL, AWS, and Airflow.
Mausam Adhikari
Associate Data Science Manager, Kathmandu
At a Glance
Open to new roles
ROI lift, Customer X360 & FCI
70%
Risk reduction via anomaly detection
25%
People led, R&D & Fraud Analytics
19
70%
Reported ROI lift from Customer X360 and Financial Credibility Index work
25%
Risk reduction through anomaly detection and AML analytics
50%
Processing speed improvement from automated ETL pipelines
19
People led across the R&D function and Fraud Analytics Scrum team
What I do best
Applied intelligence for financial systems.
My strongest work sits where model quality, data reliability, and business trust meet ‐ I build the pipelines and models myself, lead the team that ships them, and sit across the table with clients to scope and sell the solution. Research work includes a peer-reviewed paper on explainable anomaly detection (0.968 ROC-AUC, SHAP/LIME/CIU) ‐ details under Credentials. You can read more about my thoughts on data engineering and applied AI on my blog.
Credit Risk Intelligence
Building financial credibility features, NPL analysis workflows, customer segmentation, and risk-ready data marts for lending and portfolio decisions.
Data Engineering
Creating Airflow and PySpark pipelines, clean feature stores, automated reporting flows, and analytics infrastructure that teams can trust every day.
Executive Analytics
Turning complex financial signals into Tableau, Power BI, and SQL-backed decision views for leaders, analysts, and product teams.
Team Leadership & Client Delivery
Leading R&D and Fraud Analytics teams through Agile/Scrum, requirements gathering, and backlog decomposition, and owning end-to-end client implementations from discovery to production support.
Selected work
Projects with business context, not just code.
Representative work across financial data products, machine learning, AI interfaces, and analytics platforms, spanning clients in Nepal and international markets. Client names are withheld under standard financial-services NDAs.
X360
Growth & customer intelligence
Customer X360
Built a 360° customer intelligence engine spanning lead generation, segmentation, and RFM-based scoring to prioritize high-value customers. Designed churn and uplift models to protect revenue and grow average revenue per user, powered personalized recommendations validated through A/B testing, and developed distinct customer, merchant, and BFI personas to align product and marketing strategy.
Segmentation & RFM
Churn & uplift modeling
Recommendation engine
A/B testing
Persona design
PD
Credit risk
Financial Credibility Index & Credit Risk Scoring
Designed a digital credit-scoring system that assesses creditworthiness directly from transactional behavior, identifying salary, income, and expense patterns to determine each customer's ability to pay. Combined behavioral and credit-scoring algorithms to gauge willingness to pay, and modeled probability of default per customer to support underwriting and portfolio risk decisions.
Digital credit scoring
Income/expense detection
Ability & willingness to pay
Probability of default
Underwriting
Anomaly
Fraud & AML
Fraud and Behavior Anomaly Detection
Designed and implemented KYC, CDD, and EDD risk-scoring frameworks combining rule-based and model-based detection for both customer and merchant risk. Delivered real-time and batch-driven monitoring covering prescriptive alerts and preventative controls, using customer and device-level behavioral personas to strengthen fraud and AML defenses.
KYC/CDD/EDD
Rule & model-based detection
Real-time & batch monitoring
Customer/merchant scoring
Device personas
ETL
Data platforms
Automated Financial ETL Pipelines
Improved processing speed and reliability through scheduled pipelines, reusable transformations, and cleaner analytics layers.
Airflow
PySpark
AWS
AI
AI interfaces
Generative AI Customer Experience POCs
Researched and prototyped chat-based interfaces for customer support, analytics access, and knowledge retrieval.
LLMs
FastAPI
RAG concepts
GIS
Geospatial
Nepal Geo-Data Map
Interactive map visualizing Nepal-specific geographic and administrative data, built as a standalone GIS exploration tool.
GIS
Mapping
JavaScript
Selfcare
Personal project
Selfcare : Personal Finance Automation Suite
A Python automation suite for Nepali finance workflows: parses banking emails and SMS into structured transactions, deduplicates and stores them in MongoDB/MySQL, and visualizes spending through cash-flow, category, and Sankey dashboards. Also automates NEPSE price/floorsheet scraping and NEA bill checking via a registry-driven CLI.
Python
MongoDB
MySQL
Dash
NEPSE scraping
Experience
Built across analytics, engineering, and teaching.
Associate Data Science Manager
ExtensoData (F1soft Group), Kathmandu
- Manage backlog grooming, solution architecture, and research initiation, guiding R&D team members through delivery.
- Advise business and product on new areas of research; initiate and deliver proof-of-concept results to validate direction before investment.
- Meet clients directly to understand pain points, curate a solution around them, and manage it through to product.
- Support business through RFPs, client presentations, and solution demonstrations, translating requirements and managing technical risk between the R&D team and client/product expectations to deliver research products.
Team management
Client implementations
Solution architecture
RFP & presales
Agile/Scrum
Senior Data Scientist
ExtensoData (F1soft Group), Kathmandu
- Led a 7-member R&D function and a 12-member Fraud Analytics Scrum team, mentoring toward a collaborative, delivery-focused culture.
- Owned end-to-end client implementations across 4 projects: requirements discovery, data mapping, ETL design, and production support.
- Built scalable PySpark/Airflow/AWS pipelines improving processing performance by ~50%, and developed Customer X360 and Financial Credibility Index capabilities contributing to up to 70% ROI and 25% risk reduction.
- Designed anomaly detection and fraud analytics solutions strengthening AML capabilities, conducted NPL analysis for risk strategy, and led GenAI research and POCs on LLM architecture and retrieval design.
Financial analytics
Anomaly detection
AML
PySpark
Airflow
GenAI/LLMs
Data Science Instructor
Broadway Infosys, Kathmandu
- Teach Python, data science, and machine learning to 100+ students through hands-on projects and real-world datasets.
- Cover programming fundamentals, statistics, and visualization, guiding students through applied ML projects end-to-end.
Python
ML education
Visualization
Teaching Assistant, AWS and Data Analytics
Genese Cloud Academy, Kathmandu
- Developed and delivered AWS Data Analytics course content for the Advanced Digital Skills and Global Certification Program.
- Supported Python and SQL labs, and managed assignments and project evaluation.
Backend Developer
Uptechsys, Kathmandu
- Built FastAPI and Sanic-based REST APIs and services using test-driven engineering practices.
- Designed database models for portfolio management and accounting management systems.
Credentials
Education, research, and certifications.
M.Sc. in Computer Systems and Knowledge Engineering
Pulchowk Campus, Tribhuvan University
Research: Interpretable Model for Anomaly Detection in P2P Financial Transactions ‐ published in the Journal of Advanced College of Engineering and Management (Vol. 12, No. 1, 2026). XGBoost-based detection reaching 0.968 ROC-AUC with SHAP/LIME/CIU explainability analysis.
B.E. in Computer Engineering
Kathmandu Engineering College, Tribhuvan University
Early & academic projects
GhumGhaam ‐ The Journey Full of Information
Winning project at AI Expo 2019. Application that recognizes popular landmarks in Kathmandu Valley and surfaces relevant information about them.
Kids Learning Application with AI
Learning app for kids covering drawing, mathematics, and English alphabets, with custom-built drawing recognition, alphabet recognition, and handwritten number recognition models.
ChatBot Framework
Intelligent chatbot that answers questions based on provided blog links, with a scalable design supporting either AI-generated or manual responses.
Technical stack
Tools I use to move from raw data to production insight.
Modeling
Python
Pandas
NumPy
Scikit-learn
TensorFlow
Statistical modeling
Feature engineering
Data platform
SQL
PySpark
Apache Airflow
ETL/ELT
AWS S3
Lambda
Glue
Docker
Analytics delivery
Tableau
Power BI
Executive dashboards
Metric design
Stakeholder reporting
Contact
Let us build the next trustworthy data system.
Best fit: engineering/data science leadership, fraud analytics and AML, credit risk intelligence, client implementations, and applied AI prototypes. Remote, open to opportunities globally.
me@mausamadhikari.com.np
linkedin.com/in/mausamadh
github.com/mausamadh
Read my blog
@mausamadh
Based in Kathmandu, Nepal ‐ remote, open globally
Open to data science leadership, fraud/AML analytics, and AI consulting conversations, anywhere in the world.