AI/GenAI Engineer - Contract

Tanisha Systems; Hyderabad; Hybrid; INR 2,400,000 - 3,600,000; Full time

Job summary

Tanisha Systems invites an AI/GenAI Engineer to join the AI Core team in Hyderabad. You will own the design, development, and integration of AI capabilities for reimbursable cost matching, anomaly detection, and in-month billing monitoring, built on Google Vertex AI and Gemini.

Collaboration with backend and data engineering peers is essential to embed AI-driven automation in the billing pipeline. You will work on model explainability, guardrails, and integration with enterprise workflows,

Qualifications

• 4+ years hands-on experience building ML/AI systems in production.
• Proficiency with Google Vertex AI components and Gemini API.
• Strong Python skills for model dev and serving wrappers.

Responsibilities

• Design and implement reimbursable cost matching models using Vertex AI and Gemini.
• Develop anomaly detection and in-month billing monitoring capabilities.
• Build AI-driven interfaces to assist billing analysts during review.
• Define confidence scoring and model explainability outputs for auditability.
• Collaborate with Backend and Data Engineers on API contracts and pipelines.

Skills

ML/AI systems

Vertex AI

Gemini API

Python ML

GCP data services

Tools

FastAPI/Flask

Job description

AI/GenAI Engineer

• Job Title: AI/GenAI Engineer
• Experience: 4-7
• Reporting To: Solution Architect / Tech Lead
• Engagement Type: Full-time/Contract
• Location: Hyderabad

Role Overview

Seeking an AI / GenAI Engineer to join the AI Core team for the UC-003 Billing Prep program. This role owns the design, development, and integration of AI and generative AI capabilities including reimbursable cost matching, anomaly detection, and in-month billing monitoring built on Google Vertex AI and Gemini. The engineer will work closely with backend and data engineering peers to embed AI-driven automation into the billing preparation pipeline, replacing manual judgement calls with explainable, auditable model decisions.

Required Qualifications

Technical Skills

• 4+ years of hands-on experience building and deploying machine learning or AI systems in production: classification, NLP, anomaly detection, or recommendation systems preferred.
• Proficiency with Google Vertex AI: Vertex AI Workbench, Model Registry, Vertex AI Pipelines, Prediction endpoints, and Feature Store.
• Experience with Gemini API and Vertex AI Generative AI Studio: prompt engineering, structured output, grounding, and function calling.
• Strong Python skills: PyTorch, TensorFlow, or scikit-learn for model development; FastAPI or Flask for serving wrappers.
• Familiarity with GCP data services: BigQuery, Cloud Storage, Pub/Sub used as upstream data sources and downstream output sinks.

Understanding of MLOps principles: experiment tracking, model versioning, CI/CD for ML, and production monitoring.

Experience

• Prior experience building AI features in financial, billing, AP, or ERP contexts: cost classification, invoice matching, or spend analytics preferred.
• Experience integrating LLMs into enterprise workflows with appropriate guardrails, human escalation paths, and auditability.
• Demonstrated ability to iterate rapidly on model quality based on subject-matter-expert feedback within an agile delivery cadence.
• Exposure to responsible AI practices: explainability (SHAP, LIME, Gemini grounding), bias evaluation, and model documentation standards.

Preferred Qualifications

• GCP Professional Machine Learning Engineer certification.
• Experience with Vertex AI Agent Builder or LangChain-on-GCP for agentic workflow patterns.
• Background in real estate, facilities management, or professional services billing contexts.
• Familiarity with enterprise AI platform architecture.
• Knowledge of vector databases (Vertex AI Matching Engine or AlloyDB pgvector) for semantic similarity in reimbursable matching.

Key Responsibilities

AI Feature Design & Development

• Design and implement the reimbursable cost matching modelleveraging Vertex AI and Gemini to intelligently classify billing line items as reimbursable or non-reimbursable based on contractual rules, historical patterns, and contextual signals.
• Build and deploy anomaly detection capabilities to flag billing exceptions, unusual charge spikes, and data quality issues before human review.
• Develop in-month billing monitoring agents that proactively surface trends, incomplete accruals, and at-risk line items throughout the billing cycle, not just at period-end.
• Implement Gemini-powered natural language interfaces or copilot features to assist billing analysts during the human review gate (e.g., explain anomaly rationale, suggest classification, surface similar historical cases).
• Define and instrument confidence scoring and model explainability outputs so reviewers can trust and interrogate AI recommendations.

MLOps & Integration

• Manage the full model lifecycle on Vertex AI: feature engineering, training, evaluation, versioning, deployment to endpoints, and monitoring for drift and degradation.
• Integrate Vertex AI and Gemini API calls into the backend billing pipeline via well-defined service contracts, ensuring low-latency, fault-tolerant inference.
• Build feedback loops that capture reviewer accept/reject decisions and corrections to drive continuous model improvement and fine-tuning.
• Design prompt engineering strategies for Gemini LLM tasks, including few-shot examples, chain-of-thought reasoning patterns, and output schema enforcement.
• Collaborate with the Backend Engineer on API contracts and the Data Engineer on feature store design, training data pipelines, and ground-truth labelling workflows.

Quality, Safety & Governance

• Define AI evaluation frameworks and offline test suites with precision, recall, and F1 benchmarks for all classification and detection tasks.
• Implement human-in-the-loop guardrails: ensure no AI decision bypasses the review workbench and that all model outputs are logged with full provenance for audit.
• Conduct bias and fairness assessments to ensure the model does not systematically misclassify certain vendor, cost centre, or contract types.
• Document model cards and AI system design decisions to support compliance, audit, and future model handover.
• Support QA during UAT with test data generation, model stubbing, and scenario validation for edge-case billing patterns.