Amar S.

Key Skills

Software

Don't disclose

Top Subject Matter

AI-powered document (receipt/invoice) extraction and workflow automation

AI workflow automation for HR operations with structured processing

Legal Services & Contract Review

Top Data Types

Document

Image

Text

Top Task Types

Function Calling

Freelancer Overview

AI Expense Operations Agent — Full-Stack AI Engineer (2025). Brings 5+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Don't disclose. Education includes Bachelor of Technology, Delhi Technological University (2022) and Undergraduate Program, Walmart Global Tech (2024). AI-training focus includes data types such as Document and labeling workflows including Function Calling.

Labeling Experience

AI HR Ops Agent — Full-Stack AI Engineer (2025)

Don't discloseDocumentFunction Calling

Designed an AI HR operations workflow that automates onboarding/offboarding document and process handling with deterministic automation plus human approval gates. Integrated Gemini inference with model fallback to ensure robust structured extraction and workflow control across HR entities. Implemented secure role-based access control and auditing around the extracted/processed outputs to support correct downstream decisions. • Temporal-orchestrated automation with approval gates • Gemini inference with fallback for reliability • JWT authentication and RBAC for workflow access • Admin workflow controls and metrics dashboard

2025 - Present

AI Expense Operations Agent — Full-Stack AI Engineer (2025)

Don't discloseDocumentFunction Calling

Built an OCR-first document extraction workflow for WhatsApp expense reimbursement that transforms ingested images/PDFs into structured outputs for downstream validation and risk checks. Applied Gemini LLM parsing to extract fields into structured JSON, including confidence-based vision fallback for low-confidence cases. Implemented fraud/duplicate detection logic as part of the extraction pipeline to support correct labeling and approval routing. • OCR with Tesseract for initial extraction • Structured JSON extraction using Gemini • Confidence-based fallback to vision methods • Fraud/duplicate detection plus admin review and audit logs

2025 - Present

Education

Work History