SALESFORCE · AGENTFORCEHired
Refund triage agent for an e-commerce GCC
Reads inbound tickets, classifies refund-vs-replacement, executes Service Cloud flows when within policy, escalates the rest.
Every cohort ships three production projects — an LMS, HRMS, and CRM — each wired up with the AI agents you learned to build. Then the capstone deploys into a real partner org.
LMS, HRMS, CRM. Built end-to-end. Each one shipped to a staging environment recruiters can clone and run.
A full learning-management platform: course authoring, content versioning, learner tracking, assessments, and an AI tutor agent that adapts to each student's pace.
Employee data, payroll, recruitment, performance — with an onboarding agent that auto-provisions access and schedules first-week intros.
Sales, support, productivity — plus an escalation agent that reads tickets, drafts replies, and fires refund transactions when policy permits.
Pick a real partner problem. Ship an AI agent into their Salesforce or ServiceNow tenant. Walk away with a production artifact — and often, an offer letter.
Every project moves through the same six checkpoints. You demo, you commit, you ship — or you stay in the gate until you do.
One-page PRD. Acceptance criteria. Architecture sketch.
Repo, CI, auth, the boring infra you only do once.
The actual product. Pair-programmed with AI, reviewed by mentors.
Wire the AI agent in. Guardrails, eval suite, monitoring.
Live demo to a hiring partner. Defend trade-offs.
Public URL, README, screen-cap. Linked from your resume.
A sampling. Each capstone ships into a partner tenant — the link below opens the artifact, the README, and the demo video.
Reads inbound tickets, classifies refund-vs-replacement, executes Service Cloud flows when within policy, escalates the rest.
Auto-summarizes P1 incidents, drafts post-mortems, posts to Slack. Cut MTTR documentation time by 64% in week one.
Listens to Airflow + dbt failures, classifies them by severity, opens PRs against common failure modes — all on call rotation.
Reads CircleCI logs, identifies flaky vs real failures, retries the flaky ones, files a Jira on the rest. Saved 11 dev-hours/wk.
Watches CPQ quotes, applies regional pricing, flags margin breaches, drafts the customer email — never sends without sign-off.
Replaces a weekly Excel ritual. Pulls Snowflake, runs Prophet, generates the deck, posts the weekly Slack summary.
Provisions Okta + Slack + Jira, schedules first-week 1:1s, sends the welcome packet — under one human approval per hire.
Reads Figma, runs heuristic checks, posts a structured critique to the design Slack — before the human reviewer opens the file.
Reads PowerBI dashboards, narrates the week's anomalies, suggests follow-up cuts. Replaces a weekly executive memo.
Book a 20-minute advisor call. We'll walk you through a real cohort's capstone — code, demo, hiring outcome — and map it to your background.