At DataGroomr, I built the in-app AI Assistant for Salesforce data: it diagnoses data problems, explains record matches, drafts rules from plain English, and takes confirmed actions. I also built an enrichment engine that plans work from a prompt and connects to 25+ data providers through MCP.
I own the platform’s LLM infrastructure across Anthropic, OpenAI, Gemini, and AWS Bedrock, including routing, evaluation, and observability. I lead a team of seven engineers and added AI rule-building features that reduced rule-authoring support requests by roughly 90%.
At Sila Insights, I built a multi-tenant RAG pipeline and the social-data collection service, which ingested over 100 million posts and reviews. I also trained and served Arabic NLP models across 20+ dialects for BrandHealth360.
Earlier, at Neural Lines, I built a live-vision system for a Red Bull-sponsored Fortnite tournament that detected eliminations at about 99% accuracy in production, alongside document and payment-processing products. I also designed, built, launched, and marketed FilerKaro, a self-serve tax-filing product for Pakistani freelancers.
