Akarsh Mishra

About Akarsh Mishra

I am an applied AI engineer and technical leader focused on building scalable systems across LLMs, multimodal generation, and distributed infrastructure. My work centers on turning foundation models into reliable products that operate under real-world constraints.At TrueFan AI, I have led the development of distributed video infrastructure and agentic platforms, supporting high-throughput enterprise workloads for clients including AWS, Zomato, Goibibo, Agoda, Hero MotoCorp, Indomie, and Bajaj Finserv, among others. This has involved architecting large-scale inference systems, orchestrating model stacks across language, vision, and speech, and integrating acquired AI technologies into production environments.I have built multi-agent systems, real-time digital human interfaces, and end-to-end generative pipelines, with a focus on system design, inference efficiency, and asynchronous execution. Much of this work has involved managing GPU infrastructure, designing workflow orchestration, and ensuring systems remain stable under continuous load.Earlier, I worked on modeling and system design, evolving a GAN-based video system into an audio-driven lipsync model by redesigning architecture and training objectives achieving better generalization.I am particularly interested in agentic systems, multimodal interfaces, and low-latency AI, especially where strong engineering determines whether these systems are usable in practice.If you’re looking for an Engineering Manager or Leader (with a deep AI edge) who delivers across product, platform, and performance, let’s connect.

Experience

1. Head of Ai Products & Agent Systems

   TrueFan AI

   Mar 2025 — Present · Gurugram, IN

   Leading development of agentic systems and real-time digital human platforms, combining LLM reasoning, speech, and video generation for interactive enterprise use cases• Architected a multi-agent growth and GTM platform spanning SEO, lead generation, and omni-channel outreach, scaling organic impressions from 1K to 6M, increasing signups ~5x, and reducing performance marketing spend by ~90%• Orchestrated production-grade multimodal pipelines for enterprise campaigns, integrating segmentation, generative video, and proprietary lipsync systems across heterogeneous model stacks• Developed AI-driven automation systems for video translation and data normalization, removing human-in-the-loop bottlenecks and enabling scalable multi-client delivery

Education

• Alma Mater School

  XII, PCM + PhE

• Ramrao Adik Institute of Technology

  Bachelor of Engineering, Electronics and Communications Engineering

  2014 — 2018