Spotium

Overview

Spotium helps sales and procurement teams uncover opportunities and market signals by combining company/master data with domain content and AI/NLP enrichment. It’s designed for complex B2B contexts—especially mechanical engineering—where understanding machine applications and vendor ecosystems is critical. My focus was building reliable data pipelines and APIs that transform raw, multi-source inputs into searchable, actionable insights for the product UI and downstream workflows.

My Role

• Owned the backend architecture (Django + MongoDB) and data model for companies, products, events, and signals.
• Implemented ingestion/ETL jobs (Python) to unify public/partner datasets, normalize entities, and de-duplicate records.
• Integrated NLP/LLM steps for classification, entity linking, and market-topic enrichment with guardrails & audits.
• Exposed REST endpoints for discovery, enrichment, and watchlists; added role-based access control.
• Built and tuned the Angular UI modules for search, detail views, and lead routing (front-end support).
• Containerized services with Docker and shipped to Azure (App Service/Containers, Cosmos-style Mongo hosting, storage).
• Set up CI/CD, health checks, metrics, and alerting; wrote load tests to validate scaling characteristics.

Architecture & Stack

What I Shipped

Sales Discovery

• Company/application search with NLP-backed synonyms and industry terms.
• Opportunity surfacing: new installs, partner mentions, product launches, RFP-style signals. :contentReference[oaicite:2]{index=2}

Data Enrichment

• Automatic firmographic and technographic fills; confidence scoring and fallbacks.
• Topic/entity linking for machine categories and supplier ecosystems (mechanical focus). :contentReference[oaicite:3]{index=3}

Scalability & Quality

• Horizontal workers with rate-limits, retries, DLQ; back-pressure signaling.
• Data QA dashboards: coverage, anomaly flags, and manual fix workflows.