Agentic Platform Architect

Agentic Platform Architect (Founding Team / Full-time, Hands-on) Location:Hyderabad Compensation:Competitive salary + meaningful ESOP Start:ASAPAbout FN7 We’re building a lean, high-speed platform foragentic AI— a multi-tenantBrand Brain + Scout (social listening & engagement)+growth modules(ad creatives, Google PMAX, Meta, budget allocator). Our mission: help brands grow with AI-native autonomy, not dashboards. Our approach: atiny core, massive leverage surface— we ship the foundation, and the ecosystem builds on top.The Role — You’ll Be the Platform Owner You’ll own the architecture and technical design of FN7’s agentic platform — from the SDK to the orchestrator. You’ll beboth architect and builder : defining patterns, writing ADRs, mentoring engineers, and still shipping code weekly.This is not a role for a passive leader. It’s for aproduct-minded technologistwho’s been afounder, 0→1 builder, or early tech lead , and is hungry to shape the next agentic platform from the ground up.What You’ll Drive Architecture & Patterns: Evolve the agent architecture (Google ADK), module lifecycle, and micro-frontend shell. Define core design patterns and write ADRs that set the tone for the platform. Developer Platform: Build theModule SDK(templates, local dev, emulators) and CI/CD pipelines. Make it effortless for external teams to ship plug-in modules fast and safely. Data & Orchestration: Design evented backends (GCP Pub/Sub, Firestore) and containerized agent runners on Cloud Run/GKE. Keep latency sub-second where it matters. Search & Retrieval: Productionize hybrid search (Elasticsearch vector + keyword) and RAG pipelines for Brand Brain & Scout. Quality & Safety: Define evaluation scorecards, golden datasets, and guardrails for reliability and brand safety. Performance & Cost: Own profiling, SLOs, and budget control — balance speed, reliability, and efficiency like a craft. Technical Leadership: Mentor a strong Senior TL toward Head-of-Platform readiness. Drive a weekly release train and light steering cadence.What Success Looks Like (First 90 Days) External module integration time≤ 5 daysfrom hand-off to staging. Weekly release train≥ 95% pass rate . Scout/Brain latency and retrieval quality measurably improved (dashboards live). Platform SDK + docscut support load by 50%across two hackathons.You’ll Thrive Here If You… Havestarted or built something yourself— a product, a tool, a startup — that got real users and traction. Think in systems: architecture, economics, and creative constraints. Enjoy complex conversations and thrive in ambiguity. Believe execution > slides — ship, measure, evolve. Value efficiency, clarity, and momentum over process. Want to work in a place where speed, honesty, and craft matter more than headcount.Must-Have Experience 8+ years building production platforms or systems; 3+ years with cloud-native architecture. Hands-on withGCP (Vertex AI, Pub/Sub, Cloud Run/GKE) , Kubernetes, CI/CD. Strong inNode.jsandPython . Elasticsearch(vector + keyword, HNSW/KNN) in production. Firestore/Firebase data modeling, security rules, emulator-driven dev. Observability at scale (OpenTelemetry/logging/metrics). Built and shipped at least oneLLM/agent systemin production. Nice-to-Have Google ADK and multi-agent orchestration patterns. Ads domain experience (Google Ads, PMAX, Meta) or RL for budget tuning. Micro-frontend shells (React/Next) and feature-flag frameworks. BigQuery, data pipelines, experimentation frameworks.How We Work Tiny core, big surface:lean in-house team; feature expansion via paid hackathons. Quality gates:evaluation scorecards and release flags; metrics decide go/no-go. Fast loops:weekly release cadence, mid-week architecture huddles, Friday demos.Compensation & Growth Competitive base +meaningful ESOP(size based on scope/seniority). Clear path toHead of PlatformorFounding Engineer → CTOtrajectory.Interview Process (1 Week) 30-min intro (Founder) 60-min deep-dive on your past platform/agent work Practical architecture exercise (24–48h): design a plug-in agent module 60-min pair session (coding + debugging) Final culture/offer conversationApply Send yourLinkedIn/GitHuband a short note on anagentic or LLM system you’ve shipped— what worked, what didn’t — tolakshmi.ng@fn7.io