Krishna M.

Key Skills

Software

Other

Toloka

Top Subject Matter

Fraud pattern detection

marketplace abuse/risk classification

behavioural data annotation (AML/Trust & Safety).

Top Data Types

Text

Image

Document

Top Task Types

Classification

RLHF

Freelancer Overview

Risk Specialist & Marketplace Intelligence Analyst | Amazon Development Centre. Brings 11+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Other and Toloka. Education includes Bachelor of Science, St. Joseph's College of Arts and Science (2017) and Professional Diploma, Shaw Academy (2020). AI-training focus includes data types such as Text and Image and labeling workflows including Classification and RLHF.

Labeling Experience

Behavioural Data Annotation & Pattern Classification (Global Contributions)

OtherTextClassification

Contributed to 2026 global marketplace annotation work by labeling accounts and assigning enforcement-relevant categories based on behavioural and abuse indicators. Performed precision-tested assessments to ensure enforcement accuracy and prevent false enforcement across regional markets. Coordinated cross-marketplace intelligence sharing to support downstream fraud and risk operations. • IE Marketplace: labelled/classified 91 NC accounts (gibberish email + abuse signals) and achieved 100% enforcement accuracy with $3,796 documented impact. • AE & SE Marketplaces: applied multi-class classification across 84 accounts (FC processing failure vs genuine abuse) and prevented false enforcement. • ZA Marketplace: identified 247 coordinated abuse accounts with geographic clustering and banking/ASIN signature patterns; enforced within 24-hour windows. • BE Marketplace: validated PFOC at 89% precision, enforced 108/128 accounts, and supported a 53.8% DNR abuse rate reduction in one week.

2026 - Present

Risk and Intelligence Analyst - Amazon Development Centre

TolokaImageClassification

Serves as a Risk and Intelligence Analyst handling enterprise behavioral data classification, fraud pattern detection, and upstream loss prevention across multiple global marketplaces. Contributes structured feedback that improves machine-learning model signals and enhances model performance under real financial stakes. Leads governance and training initiatives for cross-team consistency, while supporting operational improvements through analytics and policy enforcement. • Delivered continuous model feedback to data science teams on incorrect signals triggered by legitimate behavior, driving AI model redesign. • Evaluated risk models and automated concession assessments, validating precision and blocking inaccurate deployments. • Built SQL-derived behavioral attributes and risk dashboards to improve observability and anomaly detection across ten markets. • Authored SOP documentation, mentored analysts, and designed SQL/Advanced Excel curriculum through Amazon learning programs.

2018 - Present

AI Model Feedback & RLHF-Equivalent Contributions | Amazon Development Centre

OtherTextRLHF

Provided continuous structured feedback to data science teams on incorrect ML model signals, functionally equivalent to RLHF training loops. Conducted expert AI output evaluation using precision testing, certification, and deployment approval/blocking decisions for risk models. Improved observability by designing SQL-derived behavioural attributes for real-time risk dashboards used by cross-functional analysts. • Delivered ongoing structured feedback driving AI model redesign to improve signal precision for legitimate human behaviour. • Certified PFOC automated concession model at 89% precision and assessed JP model at 55% precision, blocking incorrect deployment. • Designed SQL-derived behavioural attributes integrated into real-time risk dashboards for anomaly detection across 10 markets. • Led governance efforts to maintain classification quality and regulatory compliance while mentoring cohorts on data classification protocols.

2018 - Present

Risk Specialist & Marketplace Intelligence Analyst | Amazon Development Centre

OtherTextClassification

Performed structured behavioural data annotation and multi-class pattern classification across 10 global marketplaces to support upstream risk and fraud decisions. Evaluated account-level signals, determined enforcement eligibility, and ensured high inter-annotator consistency through governed data classification protocols. Delivered continuous feedback loops equivalent to RLHF workflows to improve ML signal precision and model redesign outcomes. • Labeled and classified 91 accounts (NC) using alphanumeric/gibberish email and abuse signals with documented enforcement accuracy. • Applied multi-class judgment for 84 accounts across AE/SE to distinguish processing failures from genuine customer abuse. • Detected coordinated abuse ring behavior in ZA (247 accounts) using geo clustering, banking signatures, and ASIN targeting patterns. • Evaluated 675 JP accounts for DNR-to-RR progression and blocked incorrect enforcement by precision-testing PFOC at 55%.

2018 - Present

Education

Work History