About Anoop
- Local-First Agentic AI: Designing and deploying local LLM pipelines (Llama, Qwen via Ollama) optimized for resource-constrained environments. Build autonomous agentic loops that execute locally with zero data leakage.
- Fraud Infrastructure & Audit-First Design: Engineering robust, high-performance fraud detection pipelines and transaction tracking systems built for strict auditing and compliance requirements.
- Repository Hardening & Automated Security: Integrating automated code analysis and scanning tools (CodeQL, Trivy) directly into secure development workflows and containers (Bun/Node) to identify and patch vulnerabilities autonomously.
- Full-Stack Architecture: End-to-end implementation of high-performance, low-latency codebases with a focus on absolute reliability and security compliance.
- AI & Agents: Agentic AI, Local LLMs, Ollama, Llama, Qwen-VL, Prompt Engineering.
- Security & DevSecOps: Repository Hardening, CodeQL, Trivy, Static Analysis (SAST), Fraud Infrastructure, Audit Trails.
- Backend & Runtime: Bun, Node.js, TypeScript, High-Performance Architecture, Local-First Systems.
English
Native or bilingual
Experience
- LalamoveGlobal Risk Lead Hong Kong SARTECHJuly 2020 - Today (6 years and 1 month)Hong Kong, Hong Kong SAR China• ● Restructured the anti-fraud org into a flatter model, hired backend engineers, and upskilled existing team members to make the stack more transparent and self-serve for operations.• ● Owned anti-fraud strategy and execution across model direction, controls, enforcement design, and operating cadence across markets.• ● Lifted fake/offline order detection from F1 below 50 to 92 through rules redesign, then to 95 after migrating to Python; reduced manual labeling to ~400-500 samples/month from an initial ~6,000.• ● Identified high-risk driver-user pairs at ~87% accuracy and blocked unsafe matching, contributing to a 22% drop in offline orders.• ● Built anti-abuse controls (device fingerprinting, bot detection, behavioral biometrics, OTP abuse controls), including low-frequency loyalty-signup bot detection that prevented ~$100K per quarter in leakage.• ● Improved dispute performance: sustained 0.04%-0.07% dispute rates, increased win rates from near zero to ~30%, and reduced re-appeal rates from ~7% to under 1% over two years.• ● Designed and scaled investigator AI support (triage, enrichment, policy checks, evidence drafting) with hard guardrails and human-in-the-loop review across all analysts.• ● Built SageMaker Random Cut Forest + SMOTE risk-score predictors for card binding and card authorization, replacing binary 1/0 outcomes with explainable risk scores that improved investigator decision quality.
- UberRisk Insights LeadJanuary 2018 - January 2020 (2 years)India• ● Global Process Owner for Payments Risk Insights, later India Regional Lead.• ● Investigated underground fraud networks with Threat Intelligence, including a bot network of approximately 500K accounts.• ● Helped prevent approximately $870K in potential fraud exposure (loaded but not yet utilized accounts) and identified an additional approximately $75K requiring downstream dispute handling.• ● Partnered with North America Central Operations to improve food-theft detection by shifting from static rules to GPS telemetry-driven signals.
- AAA Business Solution India Pvt LtdBusiness Development Manager IndiaJanuary 2013 - January 2018 (5 years)• ● Led strategic and go-to-market initiatives for new market penetration and client growth.• ● Modernized CRM and reporting workflows to improve execution visibility and retention performance.
Recommendations
Be the first to recommend Anoop
Help this freelancer shine by sharing your experience working together.
These freelancer profiles also match your criteria
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Education
- Bachelor of Commerce (B.Com)Osmania University2009Bachelor of Commerce (B.Com)
- CAMS (ACAMS)2020CAMS (ACAMS)