AI Technical Manager (Jorhat)

Job Role: AI Technical Manager Job Location: Bangalore Work Mode: (Hybrid) Experience : 10 years Overview:
- We are looking for an experienced AI/ML Engineer. This is a high-impact, hands-on role where you will design, build, and operationalize enterprise-grade AI and machine learning solutions that transform how we do credit, risk, compliance, fraud detection, and relationship banking.
- You will be embedded in a contemporary cloud-native AWS Bedrock environment, develop and consume MCP (Model Context Protocol) servers to extend AI agent capabilities, and use GitHub Copilot inside VS Code to write production-quality Python at speed.
- You will own the full lifecycle
- from raw data to model serving to production observability
- in a regulated financial services environment. What You'll Work On
- Intelligent Credit & Loan Underwriting Build ML models for automated credit scoring, risk-tiered loan decisioning, and covenant monitoring.
- Use Amazon Bedrock RAG pipelines to extract and summarize insights from credit memos and borrower financials. Deliver SHAP/LIME explainability layers for regulatory audit readiness.
- MCP Server Development & AI Agent Extensions Design and implement MCP servers in Python that expose internal banking APIs, loan systems, KYC data, and market data as callable tools for Bedrock AI agents.
- Integrate MCP servers into multi-agent orchestration pipelines using Amazon Bedrock Agents and LangGraph.
- Manage MCP server deployments through Azure DevOps CI/CD and ServiceNow change management.
- Conversational AI & Relationship Manager Copilots Deploy Bedrock-powered agent assistants that give Relationship Managers real-time portfolio insights, risk flags, and deal recommendations in plain language.
- Build RAG-based knowledge bases over banking policies, compliance regulations, and product documentation.
- Fraud Detection & Anomaly Intelligence Build real-time and batch ML models for transaction fraud, identity anomalies, and ACH/wire fraud detection.
- Integrate Bedrock LLMs to auto-generate analyst-ready fraud investigation summaries and connect agent tools to alert workflows via MCP.
- Document Intelligence & Extraction Design document processing pipelines using Amazon Textract, Comprehend, and Bedrock for automated extraction from KYC packages, term sheets, and financial statements.
- Expose pipelines as MCP tools for downstream agent consumption.
- Portfolio Risk & Predictive Analytics Build early warning systems for concentration risk, sector exposure, and credit deterioration.
- Develop time-series forecasting models for cash flow, deposit behavior, and interest rate sensitivity
- all instrumented with Datadog.
- AML & Regulatory Compliance AI Build ML pipelines for AML transaction monitoring and SAR pre-screening.
- Use Bedrock LLMs to draft and summarize regulatory filings and examination responses. Ensure full alignment with SR 11-7, OCC model risk standards, and BSA/AML requirements. What You'll Do Day-to-Day
- Build end-to-end ML and GenAI pipelines in Python across AWS services (Bedrock, SageMaker, Lambda, Step Functions, ECS, S3, Kinesis)
- Design, implement, and operate MCP servers
- define tool schemas, implement server handlers, version and deploy into production
- Write clean, tested, type-annotated Python using FastAPI, Pydantic, boto3, LangChain, and LangGraph
- Use GitHub Copilot daily in VS Code for code generation, test scaffolding, docstrings, and PR review acceleration
- Author and maintain CI/CD pipelines in Azure DevOps: YAML pipeline authoring, multi-stage deployments, model validation gates, and artifact management
- Instrument all AI workloads in Datadog: APM, LLM Observability (token cost, latency, hallucination monitoring), custom dashboards, SLOs, and runbook-linked alerts
- Submit all production AI deployments through ServiceNow change management: risk assessment, rollback plan, test evidence, and CAB coordination
- Own the full MLOps lifecycle: MLflow experiment tracking, model versioning, A/B testing, shadow deployment, canary releases, and drift detection
- Maintain model cards, ADRs, MCP server documentation, and data lineage artifacts for model risk governance
- contribute to the team's GitHub Copilot and AI engineering best practices Must-Haves
- 8+ years of professional experience in ML engineering, AI engineering, or applied data science in production environments
- Expert Python skills: pandas, NumPy, scikit-learn, FastAPI, Pydantic, asyncio, pytest
- Hands-on Amazon Bedrock experience: model invocation, Agents, Knowledge Bases, and Guardrails
- Working knowledge of MCP (Model Context Protocol): building and consuming MCP servers, tool schema design, and agent integration
- Proficiency with Azure DevOps CI/CD: YAML pipelines, environment gates, artifact registries, staged rollouts
- Datadog experience: APM, LLM Observability, dashboards, monitors, and SLO management
- ServiceNow change management: authoring standard/normal change requests, rollback plans, CAB submissions
- GitHub Copilot for day-to-day development within VS Code
- Familiarity with financial services regulatory frameworks: model risk management, explainability, data governance Nice-to-Haves
- Prior experience in commercial banking, FinTech, or financial services
- AWS Certified Machine Learning Specialty or AWS AI Practitioner
- Experience with LangGraph, CrewAI, or Amazon Bedrock Agents for multi-agent orchestration
- Exposure to SR 11-7, OCC Bulletin 2011-12, or NIST AI RMF Apply on Kit Job: kitjob.in/job/4mvlv9