• Currently working as an AI Engineer at Go India Advisors
• Building agentic AI systems with multi-tool calling & deep research
• Exploring advanced LLM reasoning & cost-efficient GPU infrastructure
• Available for contract-based work
• I write articles on Hashnode
• Reach me at officialkarankulx@gmail.com

Built a state-of-the-art agentic AI assistant for equity research, serving top mid-cap Indian stock exchange companies, fund managers, and individual investors. The platform performs deep research across broker reports, earnings transcripts, quarterly/annual filings, financial data, and 27+ SQL tables using multi-tool calling with the latest AI models. Implemented credit systems and monitoring with Langfuse, driving a 50% increase in paid users.

Deployed a multi-node Elasticsearch cluster from scratch as a scalable vector database with proper data backups and recovery — replacing Pinecone to optimize costs at scale.

Built multiple GPU-powered indexing pipelines, including one using MinerU to extract and crop images from raw broker reports, then analyze them in context using Gemini — achieving 95% extraction accuracy. Dynamically scaled GPU instances for task-specific processing, enabling analysts to search and integrate visual data across Indian stock companies while significantly reducing infrastructure costs.

Extensive Python scripting for schedulers, data migration pipelines, and backend APIs powering production financial systems.

• Designing a Scalable Architecture for Time-Consuming Background Processing Services

Let's connect and build something amazing together!