Role

You will wire together video, audio and image generation models into real product features that our users actually touch. This is hands-on integration work, not research. You take third-party and open models, plug them into our stack, and tune the plumbing so the output is quick, cheap and valuable enough to ship.

The right person has already done this once. You have taken a text-to-video, text-to-image, TTS or speech model and made it work inside a live product, and you know where these integrations break.

What you will do

Integrate video, audio and image generation and understanding models into our product surfaces through APIs and SDKs.

Tweak and tune integrations for quality, latency and cost: prompt formatting, parameter tuning, pre and post-processing, format conversions, chaining models together.

Handle the messy edges: aspect ratios, codecs, sample rates, frame handling, file sizes and streaming.

Build small internal tools and test harnesses to compare model outputs and pick what ships.





Work with product and backend teams to move a working prototype into production.

What we are looking for

Prior hands-on experience integrating video and image models into a real product is mandatory. You should have shipped with named models such as Stable Diffusion or SDXL, Flux, Midjourney, DALL-E or Imagen on the image side, and Runway Gen-3, Pika, Kling, Luma or similar on the video side.

One to two years of hands-on work integrating GenAI models into a live product, ideally at a rapid-moving consumer or SaaS company.

Comfortable calling model APIs and self-hosted model endpoints in Python or Node, and reading their docs to figure out the right parameters.

Practical understanding of at least one modality end to end: video, audio or imagery.

You have debugged a slow or expensive model integration and made it better.

📌 Multimodal Integration Engineer Noida
🏢 hungama digital media entertainment private
📍 Noida