Role Overview

We are seeking an exceptional, highly hands-on VP – Lead Semantic & AI Agent Engineer to join our technology team in Pune. In this critical technical-lead role, you will serve as the Platform Tech Lead for our newly established, dedicated Semantic Layer Squad.

This is a hands-on, code-first engineering role. It is not a program management or PMO role. You will be responsible for leading the technical execution, architectural alignment, and delivery of our Semantic Agent Layer—the intelligent cognitive middleware that bridges our virtualized data products with our downstream AI Consumer and Reporting ecosystems.

You will lead an agile engineering pod, run daily sprints, and collaborate closely with our core Data Services teams—specifically partnering with our Distribution Squad (who manages the underlying data virtualization, raw API exposure, and foundational MCP infrastructure) and aligning with the overall SVP Data Service Lead based in Pune. Your mission is to design, develop, and deploy autonomous Data Agents that encapsulate domain-specific business logic, rules, and governance, exposing these agents' cognitive capabilities as high-level tools via the Model Context Protocol (MCP) to enable secure, performant, and compliant Agent-to-Agent communication across our global enterprise network.

A core expectation of this role is an AI-First Thinking Mindset. You will actively adopt and champion the use of generative AI tools (such as GitHub Copilot, Claude, ChatGPT, and other developer productivity assistants) to significantly accelerate software delivery, automate testing, and improve code quality within your squad, adapting seamlessly to the enterprise-approved tools available within our network.

Key Responsibilities

1. Semantic Agent Engineering & Implementation (Hands-on Coding)

• Build Semantic Data Agents: Design, develop, test, and operate production-grade Semantic Data Agents using Google ADK, LangChain, LangGraph (or equivalent agentic frameworks). Engineer agents that encapsulate governed business meaning, analytical logic, metric definitions, policies, and domain semantics. Ensure all agent outputs are grounded in authoritative systems of record and governed semantic definitions. No unverified, non-traceable, or low-confidence information may be released to downstream agents, users, or systems without explicit Human-in-the-Loop (HIL) approval.
• Own the Semantic Data Layer: Translate complex business data concepts into reusable semantic assets, including certified metrics, KPIs, dimensions, hierarchies, calculation logic, thresholds, business rules, decision policies, and contextual definitions. Establish a single source of truth for business meaning that can be consistently consumed across agentic, analytical, and reporting workloads.
• Verification, Provenance & Trust Controls: Implement verification-by-default architectures incorporating source grounding, data lineage, confidence scoring, explainability, policy validation, and auditability. Ensure agents provide transparent reasoning, traceable evidence, and verifiable references for every business-critical conclusion, recommendation, calculation, or alert.
• MCP Tooling & Semantic Service Exposure: Consume and build MCP tools that expose governed business metrics, semantic knowledge, analytical reasoning, and decision-support capabilities. Design MCP interfaces that abstract underlying data complexity and provide trusted, business-aligned access patterns for downstream agents.
• Governance & Enterprise Readiness: Ensure semantic services can operate reliably in regulated environments requiring compliance, traceability, and reproducible outcomes.

2. Agile Squad Leadership & Sprint Execution

• Agile Tech Leadership: Act as the Technical Lead for the Semantic Layer Squad, taking hands-on responsibility for sprint planning, daily stand-ups, backlog grooming, and sprint execution.
• Code Quality & Reviews: Establish rigorous code review processes within your squad, ensuring adherence to coding standards, comprehensive unit testing (minimum 80% coverage), and robust security practices.
• Developer Mentorship: Mentor and guide junior and mid-level software developers within your squad, fostering an "AI-First" engineering culture and driving the adoption of generative AI coding assistants.

3. Cross-Squad Integration & Partnership

• Distribution Squad Collaboration: Work in close partnership with the Distribution Squad to consume their exposed data virtualization products and raw APIs, negotiate data contracts, and align schemas.
• AI Architecture Alignment: Partner closely with the wider department's AI design and Enterprise Architecture groups to ensure all conversational and agentic AI deliverables align with global enterprise standards and platform runtimes.

4. Production Estate Management & Observability

• Telemetry & Observability: Implement comprehensive logging, tracing, and metric-forwarding (using tools like Splunk, Prometheus, Grafana, or AppDynamics) within your agents and MCP servers to ensure ease of production estate management.
• Tech Debt Prevention: Proactively drive clean, modular, and well-documented codebases that prevent the incurrence of technical debt across the application lifecycle.
• Security & Compliance: Ensure all agentic workflows and tool executions strictly adhere to enterprise security guidelines, including secure token handling (OAuth 2.0 Machine-to-Machine / M2M), row/column-level masking (Apache Ranger), and secure network policies.

5. Verification, Trust & AI Risk Validation

• Source Grounding & Fact Verification Validation: Establish automated validation and evaluation frameworks (using tools like Ragas, TruLens, or custom assertion suites) to continuously test and verify that agent responses are strictly grounded in virtualized data products, eliminating hallucinations.
• Confidence & Explainability Benchmarking: Design and execute benchmarking strategies to validate confidence scoring models and ensure that the agent's reasoning paths and tool-execution logs are transparent, explainable, and easily interpretable by downstream systems.
• Lineage & Auditability Verification: Validate end-to-end data lineage and provenance tracking for every data point retrieved and decision made by the agents, ensuring the entire reasoning process is fully auditable and compliant with enterprise risk standards.
• Human-in-the-Loop (HIL) & Guardrails Enforcement: Implement and test robust guardrails (such as prompt injection defenses, PII sanitization, and output validation) and design seamless HIL handoff workflows for high-risk analytical decisions.

Technology Skills & Competencies

Required Technical Skillsets (Must be Hands-on)

• Python & AI Engineering (Expert): Expert-level proficiency in Python (FastAPI, PySpark, Flask) and hands-on mastery of Agentic AI frameworks and libraries (specifically LangChain, LangGraph, Google ADK, or equivalent) to design, architect, and orchestrate autonomous AI agents, prompt engineering pipelines, and Retrieval-Augmented Generation (RAG) architectures.
• Verification & Trust Engineering: Hands-on experience implementing and validating source grounding, fact verification, confidence scoring, provenance tracking, and Human-in-the-Loop (HIL) workflows within enterprise-grade agentic systems using modern AI evaluation frameworks (e.g., Ragas, TruLens, DeepEval, or Phoenix).
• Model Context Protocol (MCP) Integration: Proven experience integrating, consuming, and exposing tools within an enterprise MCP architecture to facilitate agent-to-agent communication.
• Database & API Design: Deep familiarity with relational databases (Oracle, SQL Server), NoSQL stores (MongoDB), advanced SQL, and designing robust, secure REST/gRPC APIs and microservices.
• AI-First Developer Skills (Must-Have): Proven capability and hands-on experience using generative AI tools (e.g., GitHub Copilot, Claude, ChatGPT, or equivalent) to accelerate software delivery, write code, generate tests, and optimize development workflows.
• CI/CD & DevOps: Hands-on experience building, configuring, and maintaining deployment pipelines in Linux environments. Strong expertise with enterprise CI/CD platforms (specifically Harness or Jenkins), containerization/orchestration (Docker, Kubernetes, and RedHat OpenShift), and enterprise workload automation/job scheduling (Autosys).

Strongly Preferred Skillsets

• Java & Spring Boot: Experience with Java, Spring Boot, and enterprise Java frameworks, as our core Data Services platforms (and the APIs exposed by the Distribution Squad) are heavily built on Java.
• Semantic Modeling & Ontologies: Strong conceptual or hands-on experience with Knowledge Graphs, metadata registries, and ontologies to define structured relationships between disparate data entities.
• Domain-Driven Design (DDD): Deep understanding of DDD principles (e.g., bounded contexts, entities, aggregates) to design clean, modular, and highly maintainable Data Agent boundaries.
• Event-Driven & Enterprise Architecture: Experience designing and implementing event-driven architectures, message queues (specifically Apache Kafka), and aligning local software designs with broader enterprise architecture standards.

Preferred / Nice-to-Have Skillsets

• Data Virtualization: Familiarity with data virtualization and query federation engines (such as Starburst, Trino, or Presto).
• Production Observability: Experience with enterprise monitoring and observability systems, specifically Splunk, Prometheus, and Grafana.

Work Experience & Qualifications

• Overall Software Engineering: Minimum 8-10+ years of progressive software engineering experience with a strong emphasis on hands-on coding, rapid feature delivery, and production launches.
• Dedicated AI Experience: Minimum 2-3+ years of dedicated professional experience focusing on AI engineering, prompt engineering, machine learning, and generative or agentic AI systems.
• Agile Team Leadership: Proven experience leading agile development teams, running sprints, and managing technical delivery for one or more squads using enterprise tracking tools (specifically Jira and Confluence).
• Education: Bachelor’s degree in computer science, Information Technology, Artificial Intelligence, or a related quantitative field.

------------------------------------------------------

Job Family Group:

Technology

------------------------------------------------------

Job Family:

Applications Development

------------------------------------------------------

Time Type:

Full time

------------------------------------------------------

Most Relevant Skills

Please see the requirements listed above.

------------------------------------------------------

Other Relevant Skills

For complementary skills, please see above and/or contact the recruiter.

------------------------------------------------------

Citi is an equal opportunity employer, and qualified candidates will receive consideration without regard to their race, color, religion, sex, sexual orientation, gender identity, national origin, disability, status as a protected veteran, or any other characteristic protected by law.

 

If you are a person with a disability and need a reasonable accommodation to use our search tools and/or apply for a career opportunity review Accessibility at Citi.

View Citi’s EEO Policy Statement and the Know Your Rights poster.