Software Engineer - AI + Video

Overview

About the Role

We are building systems that convert long-form video into high-quality short-form content.

This is not a simple clipping problem. It requires:

• Understanding dialogue, scenes, and narrative flow
• Deciding what to keep and what to remove
• Producing clips that feel coherent, complete, and watchable
• Running all of this reliably at scale

If you have tried building something like this before, you already know it is messy and difficult. Most people underestimate how hard this problem actually is.

Required Experience

• You have built or worked on a system that transforms long-form video into short-form clips

Examples include clip generators, highlight extractors, reels pipelines, or auto editors

• Or you have deep experience with video understanding, media pipelines, or AI systems applied to video

If you have not worked on problems like this, this role will not be a fit.

What You’ll Work On

AI and Video Systems

• Build systems that segment long videos into meaningful clips
• Combine transcripts, scene data, embeddings, and LLM reasoning
• Improve output quality including coherence, pacing, and completeness

Pipeline and Infrastructure

• Build and operate distributed pipelines for long-running video processing
• Work with queues, background jobs, retries, and failure handling
• Optimize cost versus latency across AI and video workloads

AI Workflows

• Use LLMs with structured outputs and multi-step pipelines
• Work with embeddings, retrieval, reranking, and evaluation

Video Processing

• Work with tools such as FFmpeg, Remotion, or similar
• Handle real-world issues such as bad inputs, codec problems, sync issues, and rendering failures

What We’re Looking For

Engineering Fundamentals

• Strong backend experience in Python or similar
• Experience with APIs, async systems, and background jobs
• Ability to debug complex systems end-to-end

AI Experience

• Hands-on experience using LLMs in real systems
• Understanding of prompt design and cost versus latency tradeoffs
• Experience generating structured outputs

Mindset

• Ability to break down ambiguous problems independently
• Willingness to work with messy real-world systems
• Focus on output quality, not just implementation

Nice to Have

• Experience with FFmpeg, Remotion, or video tooling
• Experience with Elasticsearch, vector databases, or RAG systems
• Familiarity with scene detection or speech segmentation
• Built systems that process large media files and produce real outputs

This Role Is Not For You If

• You have only built tutorial-level AI projects
• You have not worked with real-world data or production systems
• You expect fully defined tasks and specifications
• You prefer narrow, isolated backend work

About Tessact

Tessact is an AI-powered video repurposing and remix platform built for creators, agencies, and content teams.

We combine video understanding systems with generative workflows to turn long-form content into high-quality, brand-safe short-form outputs at scale.

Behind the product:

• Video pipelines that process hours of content
• AI systems operating under real cost, latency, and reliability constraints
• Infrastructure that runs in production, not just demos

Why Tessact

• Work on challenging applied AI problems in video
• Build production systems at scale
• Take ownership early and work on core systems
• Learn by solving real problems under real constraints

Final Note

If you have attempted to build systems that convert long videos into high-quality short clips and understand the challenges involved, you will likely find this role a strong fit.