AI Automation Intern

Salary Range: 20 to 25 (Currency: USD) (Pay period: per-hour-wage)

Core Responsibilities

• Workflow Orchestration (n8n): Build event-driven workflows (Slack/webhooks/cron/HTTP) and chain multiple LLM steps (router → tool calls → synthesizer; judge/fallback patterns).
• MCP Tools & Integrations: Extend MCP servers (search, ATS, audience/segments ops) with clear JSON schemas, validation, and error handling.
• Prompt Engineering (practical): Write robust system/task prompts with few-shot examples; enforce structured JSON outputs; manage cost/latency trade-offs.
• RAG & Search Hooks: Add retrieval (hybrid/BM25+vector) and citations for grounded answers; implement “no-answer” fallbacks.
• Observability & Evals: Log inputs/outputs, latency, tokens, cost, and success flags; create small golden-set evals and regression checks.
• Reliability & Ops: Add retries, backoff, dead-letter paths, Slack alerts; handle secrets, OAuth, pagination, and idempotency.
• Docs & Demos: Write concise READMEs/runbooks; present outcomes with metrics.

Special Projects (Examples)

• Slack Triage Bot: Slack trigger → LLM classifier → MCP search → LLM synthesizer → Slack reply with confidence & citations; human-in-the-loop fallback.
• Candidate Finder (ATS): MCP tool for role queries; retrieve/rank candidates; weekly adoption & quality report.
• Audience Q&A ( API): Natural-language queries → segments lookup via MCP → RAG-backed answers with source links and size estimates.
• Cost & Quality Guardrails: Judge-LLM scoring with auto-retry under threshold; per-run cost caps and daily canary tests in n8n.
• Automation Templates: Turn successful patterns into reusable n8n templates with configuration knobs and built-in metrics.

WHAT YOU'LL LEARN

• Production LLM Automation: How to ship reliable, observable LLM chains with tool-calling, schema enforcement, and fallbacks.
• Enterprise Integrations: Why MCP servers stabilize interfaces and speed adoption.
• RAG/Search in Practice: When to retrieve, how to cite, and when to say “no answer.”
• Impact Tracking: Tie automations to WAU, cycle-time saved, and $/seat ROI—and communicate results crisply.

WHO YOU ARE

• Pursuing a Bachelor’s, Master’s, or PhD in CS, Data/Information Systems, or related technical field.
• Expected graduation: [2026–2028].
• GPA 3.3+ preferred (strong projects/portfolio can substitute).
• Foundations: Comfort with at least one scripting language (JavaScript/TypeScript or Python), Git basics, and JSON/HTTP.
• APIs & Data: Willingness to learn auth, pagination, and simple SQL sanity checks.
• Quality & Logging: Basic assertions/testing mindset; interest in adding run-level logs/metrics.
• No prior expert experience required in prompt engineering or automation platforms (e.g., n8n)—you’ll get training, patterns, and mentorship.
• Automation: Exposure to n8n, Zapier, or Make.
• LLMs: Familiarity with structured prompting (JSON), tool/function calling, and context window basics.
• MCP & Tooling: Curiosity or experience with Model Context Protocol.
• RAG/Search: Embeddings, hybrid retrieval, reranking, and citation strategies.
• Observability: Langfuse (or similar) for traces, costs, and evals.
• Cloud & Apps: Basics of AWS/GCP, Slack APIs, and webhooks.
• Product Mindset: Ship small, instrumented wins quickly.
• Communication: Clear specs/acceptance criteria; concise status updates and READMEs.
• Collaboration: Comfortable pairing with PM/Eng/DS; receptive to feedback and reviews.
• Ownership: Proactive about fallbacks, edge cases, and not breaking prod.
• Learning Agility: Curious, iterative, and data-driven.
• Built at least one real automation (school, hackathon, or internship).
• Demonstrated two-LLM chaining (e.g., router → tool → synthesizer, or solver → judge).
• Hands-on with MCP-style tools, Slack bots, or lightweight RAG.
• Small eval/golden-set portfolio or open-source contributions.
• Experience showing outcomes (adoption, latency, cost) and presenting results.