Agentic AI ILT Trainer / Instructor (Mumbai)

Role Summary

We are looking for a practitioner-instructor who has shipped production agentic AI systems and can teach from lived experience. This is not a theoretical course. Participants will be expected to build and defend a production-grade system by Week 10, and the trainer must be equipped to prepare them for that bar.

The trainer is responsible for delivering the knowledge layer across all ten weeks, running engineering clinics at key architectural decision points, and providing mentorship support throughout the cohort.

Key Responsibilities

ILT Delivery (40 hours)

Deliver structured instructor-led sessions covering the full curriculum across Weeks 0–10 Ensure each session gives participants the conceptual and practical grounding they need to complete that week's milestone Use worked examples, live coding walkthroughs, and domain-grounded scenarios drawn from the insurance claims use case Build and maintain session materials (slides, notebooks, reference code) aligned to the weekly curriculum

Engineering Clinics (included in the 40 hours)

Facilitate four mandatory engineering clinics at Weeks 3, 6, 7, and 8 (30–45 minutes each) Review teams' architectural designs before implementation begins (state machines, harness design, multi-agent responsibility maps, judge calibration) Use a Socratic approach. Ask questions to surface assumptions rather than prescribe solutions Identify structural design flaws early and provide actionable correction before teams build on a faulty foundation

Mentorship (25 hours)

Provide async and/or live mentorship support to teams across the 10-week program Support teams in navigating technical blockers, architectural decisions, and milestone preparation Participate in (or observe) the three mock client interactions at Weeks 0, 5, and 10 to provide post-session coaching

Required Qualifications

Non-Negotiable

Hands-on experience building and deploying production agentic AI systems (not demos or prototypes) Proficiency with Google Cloud Platform, specifically Vertex AI, Cloud Run, Firestore, Cloud Trace, Cloud Build, and Secret Manager Practical experience with agentic frameworks, Google ADK, LangGraph, or equivalent Experience with RAG system design, including grounding, chunking strategies, retrieval evaluation, and citation-aware response design Understanding of LLM evaluation methodologies, LLM-as-judge design, judge calibration (Cohen's kappa), and EvalOps pipelines Familiarity with OWASP LLM Top 10 and AI safety/security practices including prompt injection, PII redaction, and tool poisoning Prior experience delivering technical training, workshops, or mentorship to engineering audiences

Strongly Preferred

Experience with multi-agent system design, supervisor-worker patterns, A2A handoffs, memory governance Familiarity with MCP (Model Context Protocol) for enterprise tool integration Experience with stateful workflow design, state machines, HITL gates, Firestore checkpointing Exposure to FinOps for LLM workloads, per-call cost tracking, model routing, token optimization Experience designing streaming UIs for agentic systems (SSE, Next.js or equivalent) Background in enterprise software delivery, understanding of ADRs, CI/CD pipelines, OpenAPI specs, and production deployment standards

Nice to Have

Domain familiarity with insurance claims processing (FNOL-to-settlement lifecycle, claim types, fraud signals, regulatory HITL requirements) Experience with GraphRAG, multimodal inputs, or advanced retrieval techniques

What We Are NOT Looking For

Instructors who teach from slides without hands-on delivery experience Academics or researchers without production deployment experience Generalist AI trainers without specific agentic systems or GCP depth Vendors pitching a pre-built curriculum as we have our own curriculum

Engagement Structure

Total duration: 10 weeks (target start: June 7, 2025) ILT sessions: approximately 4 hours per week, delivered in focused blocks (exact schedule to be agreed with L&D;) Engineering clinics: 4 sessions of 30–45 minutes each at Weeks 3, 6, 7, and 8 Mentorship: approximately 2.5 hours per week, delivered async and/or via live office hours Mock client interactions: trainer participation/observation at Weeks 0, 5, and 10 (post-session debrief coaching) Pre-program: 2–3 hours of curriculum alignment and session planning with L&D; before kickoff Compensation: 3 lacs

What We Provide

Detailed week-by-week curriculum with milestone specifications, supervisor checks, and engineering standards rubric Full client data package: synthetic policy documents, claim records, adjuster SOPs, historical resolved claims, and mock enterprise API specs GCP project access for session delivery and hands-on demonstrations L&D; coordination support for scheduling, logistics, and Workday tracking

This program is designed to produce engineers who can ship agentic AI to production clients. We are looking for a trainer who has done exactly that. Apply on Kit Job: kitjob.in/job/4n6cbe