Full Stack Developer - Platform & Data Systems - Immediate

About Hema AI (tryhema.com) Hema AI is building the analytics and visibility layer for the AI search era — tracking how brands appear, get cited, and get recommended across ChatGPT, Perplexity, Gemini, and other AI engines. We're pre-launch, moving fast, and looking for a senior engineer to take end-to-end technical ownership of the platform: architecture, scraping infrastructure, data systems, our agent-building dashboard, and our enterprise/white-label API layer.

The Role You'll own the core technical systems of Hema AI — large-scale scraping infrastructure (including scraping AI/LLM interfaces), our analytics platform (pipelines, storage, dashboards), our in-house RAG-powered agent builder, and the APIs powering enterprise/white-label offerings. You'll audit and harden what exists, make the architecture calls going forward, and be the senior technical voice reporting directly to the founder.

Tech Stack

• Languages/Frameworks: Python, Node.js, Express, TypeScript, React, Next.js
• State/Data (Frontend): Zustand, TanStack Query/Table, Zod
• Databases: PostgreSQL, ClickHouse, Snowflake, TimescaleDB, BigQuery, Pinecone (vector DB), Neo4j (graph, nice-to-have)
• ORM: Drizzle, Prisma
• Cloud/DevOps: AWS/GCP/Azure, Docker, container orchestration, Vercel or equivalent; Cloudflare/Akamai/Fastly for CDN & edge compute
• Real-time: WebSockets / SSE
• Observability: PostHog, Sentry
• Testing/QA: Postman, Vitest, Jest, Mocha, K6
• Security: Semgrep, Snyk/Aikido, GitGuardian, OWASP Top 10
• LLM Ops: Prompt engineering & versioning, semantic/response caching, cost monitoring & model selection, LangChain/LangGraph or custom orchestration
• Infra & Access: Job queues/workers (scraping, query fanout across LLM providers, agent execution), CI/CD, IAM, feature flagging
• PM: JIRA (smart commits), Git branching strategy

What You'll Own Scraping & Data Quality

• Build in-house scraping systems at scale for LLM outputs (ChatGPT, Perplexity, Gemini, etc.) — rate limits, rotating proxies, headless browser fleets, anti-bot evasion, session management
• Architect for horizontal scale: distributed queues, worker pools, retry/backoff, dedup, failure recovery; monitor scraper health, freshness, and silent failures
• Scrape responsibly — ToS/robots.txt awareness, flagging risk tradeoffs to the founder
• Build data quality monitoring distinct from uptime — accuracy/anomaly detection, reconciliation, and data lineage back to source

Cloud Infrastructure, DevOps & Continuity

• Own cloud infra end-to-end: provisioning, scaling, cost management, environment separation (dev/staging/prod), Docker/orchestration
• Define Git branching/release process, feature-flagged rollouts, and zero-downtime migrations (expand/contract patterns, no table locks on large tables)
• Own backup/DR across PostgreSQL, ClickHouse, and Snowflake — RPO/RTO targets, recovery drills, graceful degradation, and tiered/cold storage for aging data to manage retention cost

Internationalization & Global Delivery

• Design for i18n (externalized strings, locale-aware formatting), timezone correctness, and multi-currency billing for a global customer base
• Handle data residency/cross-border transfer requirements (GDPR localization, region-specific storage/isolation per tenant)
• Use CDN/edge delivery to keep latency low for customers outside the primary hosting region

Analytics Platform & Data Systems

• Own multi-tenant data architecture (PostgreSQL, ClickHouse, Snowflake) and vector search (Pinecone and similar)
• Build the dashboard layer (charts, tables, 50+ metrics) with real-time updates (WebSockets/SSE), virtualized tables, efficient chart rendering, pagination/pre-aggregation/materialized views/caching, and frontend performance discipline (Core Web Vitals, bundle size)
• Handle concurrency/conflict resolution on shared resources (e.g., multiple users editing the same config)
• Design ETL/ELT pipelines from scrapers → warehouse → dashboard; instrument with PostHog/Sentry

Agent Dashboard & RAG Pipeline

• Architect the agent-creation platform: prompt-to-task execution, RAG (chunking, embeddings, retrieval, re-ranking), knowledge base management, workflow/DAG orchestration, multi-tenant data model
• Build the frontend, including a visual node-based workflow builder (React Flow/Rete.js), and a library of reusable agent templates
• Integrate LLM APIs (OpenAI, Anthropic, etc.) with streaming, retries, and cost/token tracking; ensure agents can be scheduled, triggered, and monitored

Agent Safety & Guardrails

• Defend against prompt injection; enforce execution limits, timeouts, and cost ceilings to prevent runaway loops
• Build tool-call permission scoping, sandboxing/tenant isolation, kill-switches, and audit logging for agent actions — especially anything irreversible

LLM Cost, Optimization & Upstream Resilience

• Apply prompt engineering, versioning, and caching (semantic/exact-match) strategies; optimize context window usage
• Monitor and optimize LLM spend by feature/model/tenant; evaluate provider/tier tradeoffs
• Design graceful degradation for provider outages (retries, fallback models, circuit breakers); track model deprecations; maintain a lightweight dependency risk register

AI Search Product Capabilities

• Build AI crawler/bot log analytics from customer CDN/server logs (GPTBot, ClaudeBot, PerplexityBot, etc.)
• Build technical AEO audit tooling — robots.txt, llms.txt, structured data (schema.org) validation
• Design edge-based content negotiation to serve AI-optimized content to bots without changing the human-facing site
• Build sentiment analysis and competitive share-of-voice on AI-generated brand mentions, plus entity/citation-source relationship mapping (graph data model)
• Build a third-party integrations layer (CMS: WordPress/Webflow/Shopify/Contentful/Sanity; SEO tools; PM tools; BI export)

Enterprise API & White-Label Platform

• Design versioned, documented public APIs (REST/GraphQL) with a clear deprecation policy, auth (API keys/OAuth), rate limiting, and idempotency for critical operations
• Own IAM/RBAC and SSO/SAML for enterprise customer login; build audit logging (SOC 2-adjacent)
• Architect for white-labeling (branding, domains, tenant config); build reliable outbound webhooks (signing, retries, replay)
• Build billing/subscription infrastructure (Stripe), customer notifications, and data export/reporting (CSV/PDF, scheduled, API-based)
• Implement abuse prevention (rate limiting, quotas, circuit breakers per tenant) without hurting legitimate power users

Testing, Monitoring & Security

• Own unit/integration/API/load testing (Vitest/Jest/Mocha, Postman, K6) and LLM output evaluation frameworks (golden datasets, scoring, regression)
• Set up monitoring/alerting, on-call process, runbooks, and (as the platform matures) chaos engineering/fault injection
• Own SAST (Semgrep), dependency scanning, secrets management, OWASP Top 10, and periodic security audits

Compliance, Code Quality & Team

• Ensure data privacy compliance (DPDP, GDPR), data retention/deletion/export, and baseline accessibility (WCAG) awareness
• Audit the existing codebase/architecture; maintain a prioritized technical debt register; drive TDD/BDD, PR review, and code quality standards
• Own documentation (architecture, onboarding, runbooks, API docs); mentor junior/mid engineers and set technical culture as the team grows

Must-haves

• 5+ years full stack experience, 2+ years owning system/data architecture decisions; strong in Python, Node/Express, TypeScript, React/Next.js
• Hands-on web scraping at scale, including scraping/interfacing with LLM products, with ToS/robots.txt awareness
• Cloud infra, Docker/orchestration, environment & release management, backup/DR, and zero-downtime migrations for production databases
• Strong data architecture skills: ClickHouse, PostgreSQL, Drizzle, Snowflake; data quality/validation systems beyond uptime monitoring
• Real RAG and agentic workflow experience, including guardrails (prompt injection defense, execution limits, tool-call scoping, kill-switches)
• Prompt engineering, LLM caching, and demonstrated LLM cost optimization experience; experience with systems resilient to third-party API outages
• Experience building real-time, data-heavy dashboards (Zustand, TanStack) with strong frontend performance discipline
• Experience shipping public APIs with idempotency, rate limiting, and reliable webhooks; IAM/SSO integration; audit logging; multi-tenant/white-label architecture
• Experience with billing infrastructure (Stripe) and designing for a global customer base (i18n, multi-currency, timezone correctness)
• Experience parsing CDN/server logs for bot detection; familiarity with robots.txt/llms.txt/structured data standards; edge computing for content negotiation
• Experience building third-party integrations (CMS/SEO/PM tools) via OAuth; applied NLP (sentiment/entity extraction); graph data modeling
• Experience building LLM output evaluation frameworks; comfort with node-based/canvas UI (React Flow or similar)
• Testing/QA tooling (Vitest/Jest/Mocha, Postman, K6), SAST/dependency scanning (Semgrep, Snyk), TDD/BDD and rigorous PR review
• Data privacy awareness (GDPR, DPDP); strong technical documentation and communication skills; security-conscious by default
• Demonstrated ability to use AI coding tools strategically while maintaining full codebase ownership Submit profiles/resume on omika@tryhema.com