Applied AI Engineer (Agents & Automations) at Halo AI

About Us:

Halo AI is a passionate startup at the forefront of AI-powered creator marketing. We're building a vibrant community and revolutionizing the way brands connect with influencers.

Why we’re hiring

We run fast and lean, powered by AI. Your role: be the internal problem-solver who ships scrappy, reliable solutions that unblock GTM, Operations, Product, and Support. From bots and automations to

What you’ll do

Ship working systems: Design, build, and maintain services, scripts, and workflows with clear SLAs.

Automate everything: Orchestrate workflows (n8n/Make/Zapier/custom) across CRM, APIs, comms channels, and backends.

Build agentic apps: LangChain/LangGraph-based tools for enrichment, routing, triage, and outreach—with evals, guardrails, retries, and cost controls.

Scrape responsibly: Headless browsers + official APIs; rotating proxies, backoff, and anti-bot patterns (always within ToS/legal).

Connect data: Stand up light ETL/ELT to Postgres/BigQuery; implement RAG/search over internal docs; ensure freshness and observability.

Make it reliable: Logging, metrics, alerts, dashboards, and business-hours on-call for what you ship.

Document & hand off: Clear runbooks, diagrams, READMEs so others can extend your work.
 

Sample projects

Instagram lead intake → compliant DM triage: Collect, enrich, score, and route leads; auto-replies with human handoff; track deliverability.

AI outreach bot: Segment-based, personalized first touches + follow-ups; dedupe against CRM; A/B prompt + template testing.

Scrape → Clean → Sync: Playwright job to capture public business data, normalize, and sync to Postgres/Sheets with change detection + alerts.

RAG for Sales/Success: Vectorize pitch docs and prior wins; build a chat tool that drafts proposals and answers objections with citations.

Our stack

Languages: Python, TypeScript/Node.js, SQL

LLM & agents: LangChain, LangGraph, OpenAI/Anthropic APIs, RAG (FAISS/Pinecone/Weaviate), LlamaIndex (bonus)
 

Automation/orchestration: n8n, Make, Zapier, serverless cron/queues (BullMQ/Celery), Prefect/Airflow

Web & scraping: Playwright/Puppeteer, Requests/BS4, Apify, proxies, captcha-solving vendors (when compliant)

APIs & channels: Meta Graph + Messenger API (IG), WhatsApp Business API, Slack, Telegram, Twilio, SendGrid/Resend

Data & storage: Postgres/BigQuery, Redis, S3/Cloud Storage, Elastic/OpenSearch, vector DBs

Infra & DevEx: Docker, GitHub Actions, Vercel/Fly.io/Cloud Run, Terraform (bonus), Sentry/Prometheus/Grafana
 

What success looks like (90 days)

3–5 production workflows/tools live and used weekly

Idea → pilot cycle ≤ 7 days (small) or ≤ 21 days (medium)

Telemetry (logs, metrics, alerts) on everything shipped

Documented SOPs so teammates can run and extend your work

Must-haves

4–7+ years building production automations/internal tools in startups or product teams

Strong in Python + TypeScript/Node; can ship both CLI and small services

Real LLM app experience (tool use, evals, prompt versioning, cost/perf tuning)

Confident with APIs, auth, rate-limits, retries, idempotency

Hands-on with headless browsers/scraping and pragmatic about ToS/legal

Comfortable with databases, SQL, queues, cloud deploys

Strong product sense and communication; bias to action
 

Nice-to-have

WhatsApp Business API, Meta Graph (IG), Slack/Telegram bots

Vector search/RAG in production; knowledge graphs (Neo4j)

Data plumbing: Airbyte, dbt, Great Expectations
 

How we work

Small, senior, async-first. Crisp specs, quick demos, st