AI Engineer Ready to turn bold ideas into real-world impact? At Genpact, we don’t just adapt to change, we lead it. AI and digital innovation are transforming the way businesses work, and we’re at the forefront of it. Genpact’s AI Gigafactory, our industry-first accelerator, exemplifies how we scale advanced technology solutions to help global enterprises work smarter, grow faster, and transform at scale. Whether tackling complex challenges through large-scale models or agentic AI, our breakthrough solutions tackle companies’ most complex challenges. If you thrive in a fast-moving, innovation-driven environment, love building and deploying cutting-edge AI solutions, and want to push the boundaries of what’s possible, this is your moment. Genpact (NYSE: G) is an agentic and advanced technology solutions company. We leverage process intelligence and artificial intelligence to deliver measurable outcomes. With a strong partner ecosystem and decades of client trust, we provide innovative solutions that transform how businesses run. Powered by a team with an active learning mindset and client centricity at its core, we deliver lasting value for the world’s leading enterprises. Get to know us at genpact.com and on LinkedIn, YouTube, X, and Facebook. Job Description Role: AI Engineer - Forward Deployed Engineer | Advanced AI Practice | Location - Bengaluru preferred | Full Time What you will do Build AI back ends (with your Senior FDE) Write Python services in FastAPI and Pydantic that expose retrieval, extraction and agent features as documented REST APIs, with health checks and structured logs. Build data ingestion steps that parse CSV, JSON, Excel and PDF files, clean and deduplicate them, and load them into a database and a vector store, with basic data-quality checks. Implement and improve RAG pipelines: chunking, embeddings, vector search, citations of source documents, and a clear "I don't know" answer when the context is missing. Build structured extraction with Pydantic schemas and validation, choosing rules, a classical model or an LLM per field based on measured accuracy and cost. Contribute to tool-calling agents with typed tools, step limits, timeouts and a human approval step before any action that changes client data. Build the product front end Build and change React (TypeScript) interfaces that business users rely on: chat and search screens that stream answers and show source citations, document review and correction screens, and approval screens for agent actions. Build simple internal dashboards for evaluation results, usage, cost and errors. Connect front ends to back-end APIs cleanly, with loading, empty and error states, and sign-in through the client's identity provider. Build quick Streamlit or Gradio prototypes on synthetic data for early client demos, and turn the ones that work into proper web apps. Integrate Connect to client systems through REST APIs, webhooks and scheduled file exchanges, handling pagination, rate limits and retries. Design and query relational tables in PostgreSQL, SQL Server or Oracle, and note what the client's data dictionary does not explain. Work inside the client's repository, ticketing and CI, following their branching, review and change-approval process. Evaluate and prove Build and version evaluation datasets from real examples and labelled samples. Report precision, recall and F1 for extraction and classification, and retrieval and answer-quality metrics for RAG. Help wire evaluation checks and automated tests (back end and front end) into CI so that a change that makes quality worse cannot be deployed. Deploy and operate Containerise services and front ends with Docker and deploy them to a cloud sandbox, then, with your Senior FDE, to the client's environment. Add logging, metrics and traces (for example cost per request, p95 latency and error rate) to what you build. Write runbooks and hand-off notes that a colleague or the client's team can follow without asking you. Work with the client and the pod Take clear notes in client working sessions and send a three-part daily update: what changed, what is blocked, what is next. Demo your own features to business users in about ten minutes, and capture their feedback in the backlog. Write short field notes on edge cases and reusable patterns for the rest of the practice. What this role is not Not pre-sales. There is no quota and no demo-and-leave; the work starts when the client says yes. Not strategy consulting. Advice without working code is not the job. Not support. You fix root causes rather than run a ticket queue. Not project management. You change the numbers; someone else tracks them. Skillset: must have We keep this list short on purpose. If you can do most of these, please apply, even if you do not tick every box. Python: clean, readable Python 3.10+ with type hints, functions and classes, error handling, and Pydantic models. Testing and Git: writing pytest tests for your own code; using Git branches and pull requests. APIs: HTTP methods, status codes and JSON; building a small FastAPI endpoint; calling an external API with timeouts and retries. Web front end: HTML, CSS and JavaScript fundamentals; building a small React component with props, state and hooks that calls an API and handles loading and error states. SQL and data: joins, GROUP BY and CTEs; designing a simple table with primary and foreign keys; cleaning messy CSV or JSON with pandas. Containers and deployment: writing a Dockerfile for a Python app; you have deployed at least one project somewhere others could use it. LLM basics: tokens, context windows and cost; prompting for structured output; you have built at least one simple RAG pipeline (any vector store) with citations. ML and evaluation basics: train, validation and test splits and data leakage; precision, recall, F1 and the confusion matrix, and why accuracy misleads on imbalanced data. Security hygiene: never putting secrets in code or logs; basic awareness of personal data. Communication: explaining a technical trade-off simply in English, in writing and in person. Full stack engineering skills FDEs ship complete products, so you will work across the stack. The first list is what we test at interview; the second is what you will reach in your first 3 to 6 months with training and mentoring. Must have Front end: HTML, CSS, JavaScript (ES6+); React components, props, state and hooks. AI user interfaces: Calling an API from the browser and rendering the result. Back end: FastAPI endpoints with Pydantic validation. Databases: Simple schema design with keys; SQL joins and CTEs. Authentication: Understanding sessions versus tokens. Web security: Never trusting user input; no secrets in front-end code. Testing: Unit tests in pytest. Quality and accessibility: Readable, consistent code. Delivery: Dockerfile for a Python app. Learn in your first 3 to 6 months Front end: TypeScript in strict mode; Next.js or Vite; form handling and validation; state management (React Query or similar); a component library and Tailwind CSS; responsive layouts. AI user interfaces: Streaming LLM responses with Server-Sent Events or WebSockets; showing citations and source previews; feedback buttons (thumbs up/down) wired to evaluation data; human-in-the-loop review and approval screens. Back end: API design and versioning, pagination, background jobs (Celery, RQ or a cloud queue), caching with Redis, async I/O, basic Node.js for reading client code. Databases: Normalisation versus denormalisation, indexes, migrations with Alembic, an ORM (SQLAlchemy), window functions, pgvector alongside a dedicated vector store. Authentication: OAuth 2.0 and OpenID Connect sign-in (Azure AD / Entra ID, Okta), JWT handling, role-based access control, CORS configuration. Web security: OWASP Top 10 (XSS, CSRF, SQL injection, broken access control), secure headers, OWASP Top 10 for LLM applications, safe rendering of model output. Testing: Front-end tests with Vitest or Jest and React Testing Library; end-to-end tests with Playwright; API contract tests. Quality and accessibility: ESLint, Prettier, Ruff; basic accessibility (WCAG: keyboard navigation, labels, contrast); front-end performance basics. Delivery: Building and serving a front end in Docker; Docker Compose for the full stack; CI/CD pipelines covering front and back end; static hosting or a container service in Azure or AWS. What you will learn in your first 3 to 6 months Beyond the full stack skills above, you do not need these to apply. Our training programme and your mentor will get you there. Engineering depth: mypy or pyright in CI, packaging with pyproject.toml, mocking LLM calls for deterministic tests, Linux diagnosis and shell scripting. Enterprise integration: secret stores (Azure Key Vault or AWS Secrets Manager), webhook signature verification, rate limits and idempotency keys, GraphQL, legacy XML/SOAP, and platform APIs such as Salesforce, ServiceNow or SAP. Data engineering: reading unfamiliar enterprise schemas, scanned PDFs and OCR, incremental and idempotent pipelines with quality checks that fail the run. Cloud and operations: one cloud in depth (Azure preferred), multi-stage and non-root images, least-privilege identities, deployment with approval and rollback, reading Terraform or Bicep, dashboards and alerts. Advanced retrieval and agents: hybrid search with Reciprocal Rank Fusion, reranking, multi-turn RAG, a provider-agnostic LLM layer over at least two vendors, LangGraph agents with checkpoints and approval gates, MCP servers. Evaluation and statistics: Precision@K, recall@K, MRR, groundedness, RAGAS or DeepEval (and their failure modes), CI quality gates, sample size and confidence intervals, comparing two systems fairly, MLflow. Security and responsible AI: Genpact's four-tier data classification, prompt-injection defence, PII masking with Presidio, guardrails, audit logging, and what DPDP, GDPR, HIPAA, SOC 2 and ISO 27001 change in a design; communicating hallucination and bias risks honestly to clients. Good to have A public portfolio: a deployed full stack RAG or agent app with an evaluation harness and a README counts for more than any certificate. Internships, hackathons and open-source contributions that reached real users count too. Exposure to finance, insurance, healthcare or supply-chain documents such as invoices, claims, policies or purchase orders. Reading-level Java or C#; basic UI/UX sense (wireframing in Figma); awareness of knowledge graphs (Neo4j) or vision-language models for documents. Eligibility A minimum Bachelor's degree in computer science, engineering, mathematics, statistics or a related field, or equivalent demonstrated engineering experience. A Master's degree in above fields or data or AI is welcome. Qualifications Certifications Required Skills AI/ML Ops Language English (Required) Language Proficiency - Proficient - C2 Additional Job Location - Job Type Regular Master Skill List - Advanced Analytics / AI / ML Remote Type - Hybrid Work Shift - Day Job (India) Why join Genpact? • Lead AI-powered transformation – Drive innovation and solve real-world business challenges that matter • Make an impact – Help global enterprises solve business challenges that matter • Accelerate your career – Gain hands-on experience, mentorship, and world-class learning opportunities to stay ahead • Work with the best – Join 140,000+ bold thinkers and problem-solvers who push boundaries every day • Thrive in a values-driven culture – Our courage, curiosity, and incisiveness - built on a foundation of integrity and inclusion - allow your ideas to fuel progress Come join the 140,000+ coders, tech shapers, and growth makers at Genpact and take your career in the only direction that matters: Up. Let’s build tomorrow together. Genpact is an Equal Opportunity Employer and considers applicants for all positions without regard to race, color, religion or belief, sex, age, national origin, citizenship status, marital status, military/veteran status, genetic information, sexual orientation, gender identity, physical or mental disability or any other characteristic protected by applicable laws. Genpact is committed to creating a dynamic work environment that values respect and integrity, customer focus, and innovation. Furthermore, please do note that Genpact does not charge fees to process job applications and applicants are not required to pay to participate in our hiring process in any other way. Examples of such scams include purchasing a 'starter kit,' paying to apply, or purchasing equipment or training. We're an agentic and advanced technology solutions company that runs mission-critical operations for the world's leading companies, while building and deploying the agentic products and technology that power those operations. At Genpact, you'll start your career inside the work — with real data, real challenges, and the experiences that shape what you're capable of. Be on it because you'll be in it. At Genpact, your expertise grows because you're inside the work. Whether you build AI systems or bring deep knowledge of how businesses run, you'll work on real challenges that change how the world's leading companies operate. There's a lot to learn here, and room to grow. Genpact never asks applicants to pay to be part of our hiring process. If you get an email asking you to purchase a starter kit, equipment or training, or pay to apply for a job, please assume this is a scam.

AI Engineer 4A.

more at Genpact.

Campus Track – Consulting, Sales & Advisory

Sr. Developer - Application Development JAVA N 4B

Analyst - Service Desk 5B

similar roles.

Data Science Engineer Internship

Data Engineering Internship

Graduate Program – Application Consulting

Project Specialist

Associate, Data Science

Associate, Data Scientist