Harsh Rastogi in a bright, minimal studio office, sitting on the edge of a desk with two laptops

    Harsh RastogiAI Product
    Engineer

    atModelia

    I build enterprise-grade AI products and agents, from the first sketch to production. Then I open‑source the skills and tools that got me there.

    Enterprise AI that has to work: video, billing, support and hiring.

    At Modelia I'm the AI Product Engineer and Business AI Head. It's a complete AI solution for the fashion ecosystem: DAM and PIM, catalogue management, product imagery and visual master generators, with the extra power of bulk generation across an entire catalogue at once.

    Every week I bring a new idea and build it end to end, from the first prototype to a feature people test in production, like the AI video pipeline, the metering engine under every generation and our customer-support agents. I also look for where AI can grow the business and automate how our teams work.

    Outside Modelia I'm a founding engineer at SelfAgentic, a multi-agent AI orchestration platform now in closed beta. Before that I built AI hiring automation at Asynq, a university LMS for 5K+ learners at CCS University, and did a defence internship at Bharat Electronics Limited.

    fewer client escalations after unifying and restructuring the Modelia app for Shopify merchants
    95%+
    AI images and videos generated every day on Modelia's platform
    1000s
    learners on a university platform I built solo
    5K+
    AI products running in production
    4

    Products I've built and shipped to production.

    All work
    modelia.ai
    Modelia: The AI solution for the fashion ecosystem

    AI Product Engineer & Business AI Head

    ModeliaThe AI solution for the fashion ecosystem

    DAM and PIM, catalogue management, product imagery and visual master generators in one platform, with bulk generation across an entire catalogue at once.

    fewer client escalations after unifying the Shopify app
    95%+
    AI images and videos generated every day
    1000s
    Carcode on a Toyota's centre screen at night, answering a spoken request

    Carcode: code from your car.

    I built Carcode to turn Claude Code into a voice assistant for the drive. Your Mac does the work; you just talk.

    • Say "Hey Siri, Jarvis" and Claude Code answers out loud over CarPlay
    • Reviews PRs, catches you up on Slack, ships draft PRs on your Mac
    • No app, no API key. Installs in about two minutes
    $npx carcode-cli

    Four roles in two years, each a step up.

    From a defence internship in 2024 to AI Product Engineer and Business AI Head at Modelia in 2026, with a promotion inside five months. Each step came with more ownership: from building interfaces, to whole platforms, to deciding what the business builds next.

    1. Jan 2026 — Present10 mos · 2 roles

      Modelia

      Remote

      1. Jun 2026 — Present · Now

        AI Product Engineer & Business AI Head

        Every week I bring a new AI idea and build it end to end into a feature people test in production, across Modelia's fashion platform: DAM and PIM, catalogue management, product imagery and visual master generators with bulk generation. I also find where AI can grow the business and automate how our teams work.

        +What I built
        • Built the AI video production tool for commercial-ready content: multi-clip generation with music
        • Unified two isolated platforms into one synchronised system with cross-platform payments and shared generation history
        • Replaced rigid sprint releases with continuous delivery behind feature flags, enabling controlled rollouts and A/B tests
        • Built per-tool consumption analytics and cut AI infrastructure costs with caching and usage monitoring
        new ideas taken from prototype to production
        Weekly
        single-output generation grown into a full creative pipeline
        1 → many

        ↑ Promoted after 5 months

      2. Jan 2026 — May 2026 · 5 mos

        Full Stack Engineer

        Engineered the core platform: the generative image engine, the merchant-facing Shopify app, the Studio creative tool and internal systems serving hundreds of merchants generating thousands of AI images a day.

        +What I built
        • Secure media delivery through an Azure Blob proxy and cross-platform credit sync with PostgreSQL and Prisma
        • Real-time updates over SSE with Redis pub/sub for horizontal scaling
        • Production monitoring, structured logging and security hardening across the platform
        fewer client escalations after unifying and restructuring the Shopify app
        95%+
        to rebuild the legacy platform as a modular, production-grade system
        35 days
    2. Jan 2025 — Dec 20251 yr

      Asynq

      Gurugram, India

      1. Jan 2025 — Dec 2025 · 1 yr

        Full-Stack Developer

        Built the AI hiring automation platform from the ground up: candidate shortlisting, resume analysis at scale and an autonomous AI interviewer.

        +What I built
        • Candidate scoring and ranking engine that surfaces best-fit profiles and cuts manual screening time
        • Agentic AI interviewer that runs structured interviews with context-aware follow-up questions
        • Recruiter-first UX so non-technical hiring teams could use every AI tool on day one
        of resumes analysed per batch through async job queues
        100s
        API responses with a type-safe tRPC backend and caching
        <1s
    3. Jul 2024 — Aug 20242 mos
      1. Jul 2024 — Aug 2024 · 2 mos

        Software Engineering Intern

        Defence internship at India's premier defence electronics PSU, building interfaces for Harvester HCI virtualised computing and storage.

        +What I built
        • React and Node.js interfaces for defence-grade virtualised computing and storage
        • Worked with cross-functional defence engineering teams on mission-critical systems
        less paperwork on Indian Air Force project interfaces
        75%
        more efficient interfaces through component architecture work
        45%

    Take what you need.

    The Claude skills I use at work, the tools I open-source, and what I learn shipping AI to production. Each skill is a plain SKILL.md: read it first, then install it with one command.

    Skills install to ~/.claude/skills/<name>/SKILL.md

    From the model call to the invoice.

    More about me
    • Generative and agentic AI

      Image and video pipelines, multi-agent systems and LLM features, with evaluation and guardrails designed in from the start.

    • AI for the business

      Finding where AI earns its place, then shipping it with the metering, pricing and rollout that make it pay for itself.

    • Full-stack product engineering

      APIs, queues, data models and interfaces in TypeScript that hold up under production load.

    • Platforms and infrastructure

      Merchant apps, internal tools and cloud deployments, taken from architecture to running in production.

    AI
    • Generative AI
    • Agentic AI
    • LLM integration
    • Claude Code
    • MCP
    • Evaluation
    Languages
    • TypeScript
    • JavaScript
    • Python
    • Java
    • SQL
    Frontend
    • React
    • Remix
    • Next.js
    • Tailwind CSS
    • Shopify Polaris
    Backend
    • Node.js
    • Express
    • tRPC
    • GraphQL
    • BullMQ
    • Prisma
    Data
    • PostgreSQL
    • MongoDB
    • Redis
    • Supabase
    • Firebase
    Cloud
    • Azure
    • AWS
    • Docker
    • Vercel
    • GitHub Actions
    • Nginx

    Questions people ask.

    Who is Harsh Rastogi?

    Harsh Rastogi is the AI Product Engineer and Business AI Head at Modelia, a complete AI solution for the fashion ecosystem: DAM and PIM, catalogue management, product imagery and visual master generators, with bulk generation. He builds AI products and agents end to end, is a founding engineer at SelfAgentic, a multi-agent AI orchestration platform, and created carcode, an open-source way to code from your car with Claude Code.

    What does an AI Product Engineer do?

    An AI Product Engineer owns an AI feature from the product decision to production: choosing where a model belongs, building the pipeline and the guardrails around it, metering usage, evaluating quality, and shipping it to real users. At Modelia, Harsh does this across the AI video pipeline, the metering engine behind every generation and the customer-support agents.

    What does Harsh Rastogi do on the Modelia AI team?

    On the Modelia AI team, Harsh is the AI Product Engineer and Business AI Head. He works on fashion AI in production: bulk generation of product imagery and visual masters across whole catalogues, the AI video pipeline and the metering engine. He also built the Modelia Shopify app, which he unified and restructured to cut client escalations by more than 95%.

    What has Harsh Rastogi built?

    Modelia's AI video pipeline, metering engine and customer-support agents; Tutively, a smart school ERP with AI student analysis that he owns end to end; SelfAgentic as its founding engineer; an AI hiring automation platform at Asynq; GyanSathi, a university learning platform for 5,000+ learners that he built solo; and carcode.

    What is carcode?

    carcode is an open-source tool for coding from your car with Claude Code. You say "Hey Siri, Jarvis" and Claude Code reviews pull requests, reads Slack and ships code changes on your Mac, while Apple CarPlay reads the answers aloud. It is MIT licensed and needs no API key.

    What are Harsh's Claude skills?

    They are SKILL.md files for Claude Code drawn from production work: LLM cost audits, agentic error recovery, AI feature rollouts, Shopify app review and Prisma with PostgreSQL performance. Each one installs into ~/.claude/skills with a single command from harshrastogi.tech/skills.

    How can I work with Harsh?

    Email harshrastogi0603@gmail.com, use the contact page or connect on LinkedIn. He is open to conversations about AI products, agentic systems and enterprise-grade AI.

    Building something with AI? Let's talk.

    I'm open to conversations about AI products, agentic systems and enterprise-grade AI. Email reaches me directly.

    Email

    harshrastogi0603@gmail.com