WORK WITH SHEERSH TIWARI

AI workflows built around the work.

I build AI workflows and internal tools for teams dealing with repeated research, disconnected information, and manual handoffs. I’m Sheersh Tiwari, an independent engineer in Indore, India, working remotely.

AI workflows & internal toolsHuman review and controlIndependent engineering / Remote

Start with one useful workflow

A good first brief names the repeated task, the information it needs, and the person who checks the result. We identify the slow or unreliable step before choosing a model or adding agents.

Research gathered into a reviewable workspace.
An assistant connected to a defined set of information and tools.
A task interface that makes status, errors, and human decisions visible.

What I can deliver

The scope can include model integration, backend endpoints, a usable interface, review states, and documentation. We agree on the acceptance criteria before expanding the workflow.

A focused first version around a real task.
Explicit loading, unavailable, and failure states.
Review or approval points where a person needs to decide.
Configuration notes and a practical handover.

Relevant work: कामCREW

I built कामCREW, a Next.js and TypeScript control center around the Hermes agent runtime. It brings tasks, conversations, model configuration, and research into one workspace with pause and resume controls.

Hermes supplies the agent runtime; my work is the control layer and its integration. This is an independent build, not a claim of commercial customers or measured business savings.

Read the कामCREW case study

How an engagement starts

Share the task you repeat, the systems involved, and any data-access constraints. We define a small usable version, review it against the brief, and decide what to refine. DeepSeek, vector-database retrieval, MCP, and agent evaluations are current exploration areas; they are scoped separately from delivered capabilities.

BEFORE WE START

Common questions.

Do I need multiple AI agents?

Not necessarily. We start with the task and use the simplest workflow that meets the brief. A model integration or a conventional automation may be sufficient.

How do you handle private business information?

Before implementation, we agree on what information is needed, who can access it, and which providers may process it. The integration and hosting choices follow those requirements; no particular compliance certification is implied.

What does an AI workflow cost?

Pricing depends on the integrations, interface, hosting, review requirements, and acceptance criteria. Send a description of the workflow for a scoped discussion; there is no fixed price or delivery promise on this page.

A USEFUL FIRST STEP

Tell me what needs
to work better.

Share the task, the systems involved, and the outcome you need. We can start with a focused conversation.

er.sheershtiwari@gmail.com