Simplismart careers

There are 4 major bottlenecks for enterprises to adopt Generative AI -

1. Standardization- There is a need to standardize the workflow orchestration of fine-tuning a model to deploying it and finally observing it in production. Enterprises are planning to build these orchestration layers in-house (imagine pre/post Databricks era for ETL pipelines).
2. Cost- A major concern for enterprises at this point is to justify the RoI for the use-cases(reduce costs/ adding a revenue stream) they want to adopt GenAI for.
3. Data Privacy- Enterprises are concerned about data going out of their cloud, more so with industries like BFSI and Healthcare.
4. Control- Enterprises need precise control over a robust infrastructure to justify their SLAs and want to get rid of constraints like downtime and rate-limits that come with 3rd party APIs.

Simplismart is a cloud and model-agnostic MLOps workflow orchestration platform that helps organizations fine-tune, deploy, and observe models at scale using a declarative standardized language similar to Terraform.

We are focusing on speeding up GenAI models like text-to-image (SDXL), speech-to-text, and open-source LLMs. There is a need to solve the tooling layer to tackle problems like grid-search for most optimal serving performance, multi-tenancy LoRA serving and fast/ efficient scaling.

Simplismart is the fastest inference engine on Nvidia GPUs, supporting upto 280+ tokens/ sec. with Llama3-8B on a single H100 machine.