Mushaim Khan

Designed and tuned the AI chatbots conversational behavior — defined its response preferences,
tone, and guardrails (what it should and shouldnt say, and how it should answer) through prompt
engineering and iterative testing — improving response consistency and keeping replies on-topic and accurate (~90% on-intent responses).
Built multi-agent LLM workflows (role-separated agents: drafting → peer-review → reporting) that hand off structured output automatically, improving reliability and removing manual hand-offs.
Engineered Python automation and scheduled data pipelines integrating multiple REST APIs into a single production flow, with auth, pagination, and retry/error handling for reliable unattended runs (near 100% successful automated runs).
Owned the automation stack end to end — ingestion, processing/flagging logic, and delivery — cutting repetitive manual review for the team by an estimated ~70-80% on automated tasks.

Built scheduled Python pipelines that ingest live REST-API data, apply segmentation/anomaly logic, and auto-generate reports — cutting manual operational review by ~10 hours/week per client.
Engineered an affiliate-tracking + fraud-detection system with explainable flagging and fully automated monthly reporting — saving ~6 hours/week of manual review and surfacing risky activity that was previously missed.
Built full-stack web applications and AI integrations end to end (frontend, REST APIs, deployment) across multiple client projects.
Designed QA frameworks output-consistency edge-case coverage to models.