Sridharan S

Data Scientist

Chennai, Tamil Nadu, India5 yrs 5 mos experience

AI EnabledHighly Stable

Key Highlights

Saved over $40M through advanced fraud detection systems.
Expert in Graph Data Science and Machine Learning.
Proven ability to optimize fraud detection models.

Stackforce AI infers this person is a Data Scientist specializing in Telecom fraud detection and Graph Data Science.

Contact

Skills

Core Skills

Data ScienceGraph Data SciencePredictive Machine Learning

Other Skills

Python (Programming Language)PythonSQLNeo4jData EngineeringETLFeature EngineeringAngularSpring BootAzure DatabricksCypher Query LanguagePredictive AnalyticsExtract, Transform, Load (ETL)Oracle AI servicesOracle ML services

About

At Verizon, I built advanced fraud detection systems that have saved over $40M+ in incremental savings for a major US telecom. I am a Data Scientist with over 5 years of deep expertise in architecting and deploying end-to-end solutions in Graph Data Science and Machine Learning. I specialize in identifying and preventing sophisticated, large-scale fraud by turning massive, high-velocity data into actionable, real-time intelligence. My core competencies include: Graph Data Science & Analytics: Architecting and engineering the firm's primary fraud detection models in Neo4j. I apply a range of GDS algorithms (WCC, Degree Centrality, Node Embeddings, Node Classification) to uncover hidden networks and score entity risk. Predictive Machine Learning: Designing, building, and tuning ML models (e.g., Logistic Regression, Random Forest, XGBoost) to accurately classify high-risk transactions. Data Engineering & Pipelines: Engineering and scaling robust ETL pipelines (Luigi workflows) to ingest, process, and model data from Oracle databases. Model Optimization & Analysis: Performing advanced feature engineering and rigorous False Positive / True Positive (FP/TP) analysis using Python (Pandas, Scikit-learn). I also build Decision Tree models for deep-dive rule performance analysis and optimization. Operational Deployment: Owning the critical, high-availability system that translates graph-detected insights into negative lists, which are used to proactively stop high-risk transactions across all company channels. I am driven by solving complex data puzzles and building automated, intelligent systems that deliver a clear and measurable impact on the bottom line.

Experience

5 yrs 5 mos

Total Experience

5 yrs 2 mos

Average Tenure

3 mos

Current Experience

Feg

Data Scientist

Jan 2026 – Present · 3 mos · Hyderabad, Telangana, India · Hybrid

Python (Programming Language)Data Science

Verizon

4 roles

Engr 2 - Data Science

Jul 2022 – Dec 2025 · 3 yrs 5 mos · Hybrid

PythonSQLData Science

Data Scientist

Promoted

Mar 2021 – Dec 2025 · 4 yrs 9 mos · Hybrid

Driving $40M+ in incremental savings by architecting and deploying advanced fraud detection systems. A Data Scientist with a deep expertise in Graph Analytics (Neo4j) and Machine Learning, specializing in identifying and preventing sophisticated, large-scale fraud/losses within the telecom industry.
Key Responsibilities & Achievements:
Architected and engineered the firm's primary Graph Analytics models for correlated fraud detection using Neo4j.
Engineered and scaled robust ETL pipelines (Luigi workflows) to ingest, process, and model data from Oracle databases into the graph environment.
Leveraged a variety of Graph Data Science (GDS) algorithms to uncover hidden fraud networks, including WCC (Weakly Connected Components) and Degree Centrality to identify clusters, and applied Node Embeddings and Node Classification models for advanced entity analysis and risk scoring.
Owned the system that translates graph-detected fraud clusters into real-time negative lists, a critical component that proactively flags and stops high-risk transactions across both online and in-store channels.
Designed, built, and tuned predictive Machine Learning models—including Logistic Regression, Random Forest, and XGBoost—to accurately classify high-risk transactions, significantly improving fraud capture rates.
Utilized Python (Pandas, Scikit-learn) for advanced feature engineering, model tuning, and rigorous False Positive / True Positive (FP/TP) analysis to optimize model thresholds, balancing strong fraud defense with a seamless customer experience.
Built Decision Tree models to perform deep-dive analysis on existing SQL-based rule performance, identifying key drivers of fraud and providing data-driven recommendations for rule optimization.