An application made for this job — a tailored resume and cover letter that speak straight to the posting.
MIDCAI CONSULTING PRIVATE LIMITED is seeking a software engineer to build Claude-based agents and workflows. You will connect Claude to Salesforce, Slack, Google Drive, CRMs, and MCP servers, design prompts, and ensure the tools run reliably in production.
You will write Claude code setups, perform thorough testing, and monitor post-launch performance, cost, and accuracy, sharing learnings through demos and write-ups for the Salesforce team.
MIDCAI CONSULTING PRIVATE LIMITED | Full time
At MIDCAI, we turn complexity into momentum for high-growth companies, founders, and product leaders. We're not your typical outsourcing vendor - we're the thinking partner you call when you need to move fast without breaking things. We've built Salesforce businesses from the ground up, not just delivered projects. Whether you're scaling your Salesforce practice, automating with AI, or modernising data operations, we bring strategic thinking and hands‑on execution.
We work with teams who want more than a vendor. You want partners who understand the stakes, move at your pace, and stay focused on outcomes that matter. Let's build something lasting together.
You'll build things with Claude that people actually use. Some will be for clients, usually sitting next to their technology setup. Some will be for us: the tools our own team uses to write proposals, prepare for discovery calls, review code and get through the day faster.
Building agents and workflows on the Claude API. That includes tool use, prompt design, and making sure the agent does the right thing when the input is weird.
Connecting Claude to where the work lives: Salesforce, Slack, Google Drive, CRMs, and whatever systems a client runs on. MCP servers will be a big part of this.
Writing Skills and Claude Code setups that our team relies on daily, and improving them when people tell you what's annoying.
Testing things properly. That means writing evals, checking results, and telling us straight when something isn't ready to ship.
Keeping an eye on things after launch: cost, speed, and the cases where it gets things wrong.
Showing the rest of the team what you've learned, through short demos, quick write-ups, or pairing with our Salesforce developers.
You've built something with an LLM and then spent twice as long making it reliable, and you found that part interesting.
You read Anthropic's release notes the day they come out and want to try the new thing that evening.
You'd rather show a working prototype than write a document about one.
You're happy to explain your work to a non-technical client in plain words.
You need a detailed ticket before you start.
You think the job is mostly writing clever prompts. Prompts are maybe 20% of it; the rest is plumbing, testing and handling edge cases.
You'd prefer never to speak to a client.
3+ years writing production software in Python or TypeScript.
At least one thing you've built on an LLM API that real people used. Side projects count if people used them.
A working understanding of tool use, retrieval, evals and agent patterns. If you've built an MCP server, even better.
Comfort with APIs, Git and deploying your own work.
Clear written English. A lot of this job is explaining what the AI did and why.
Experience with Claude Code, the Claude Agent SDK, or Skills.
Any Salesforce exposure: Apex, Agentforce, Data Cloud.
Something public you've built or written that we can look at.
Where you studied, and whether your last job title had "AI" in it. Show us what you've built.