AI/Automation ConsultingthatSpeaks Business

Built a comprehensive automation platform for a talent acquisition firm struggling with manual candidate screening. The client had no CRM system - roles were scattered across chat history and notebooks, and recruiters spent hours manually browsing LinkedIn-equivalent sites (Liepin), clicking through profiles one by one.

The system I built handles the entire pipeline: accepts search criteria, automatically navigates and filters candidates, opens profiles in batches, then uses LLM with custom prompt engineering to evaluate each candidate against up to two filter criteria simultaneously. Each evaluation includes detailed scoring and reasoning.

Added an optional human review interface where recruiters can see LLM scores and reasoning before deciding who to contact. Once approved, the automation opens selected candidate URLs and sends contact messages with intelligent delays to avoid anti-bot detection. The entire stack spans 20,000+ lines across Python, MobX (UI), LLM integration, Playwright for web automation, and database architecture.

Impact: Generated $50K+ ROI in first month ($20K contract payment + $10K commission from a successful hire made during testing). Improved efficiency by >50% - processing ~20 candidates/minute vs ~5 manually, with significantly better filtering accuracy than human pattern matching.

Developed a mission-critical Excel-based pricing platform backed by Python, VBA, and quantitative modeling for S&P Global's Collateralized Loan Obligation (CLO) business. This was an internal project during my time at S&P, replacing a legacy system that had been the standard for over 5 years.

The platform handles complex CLO deal pricing calculations and became the primary tool for 30%+ of daily operations across all CLO teams. Given the high-stakes nature of structured finance pricing, the system required rigorous validation, accuracy guarantees, and seamless integration with existing workflows.

Delivered measurable 20% efficiency improvement through better data flow, automated validation checks, and improved calculation speed. The platform directly supports millions of dollars in annual revenue by enabling faster, more accurate pricing decisions for CLO transactions.

While I can't share proprietary details, this project demonstrated my ability to build financial tools for enterprise environments where accuracy, reliability, and regulatory compliance are non-negotiable.

Built an Excel-based data analytics script for a hedge fund client to streamline their IPO stock selection process. The fund needed to screen thousands of tender offers against multiple custom criteria to identify promising IPO opportunities - a tedious manual process that took 30+ minutes per analysis session.

The solution automated the entire screening workflow with intelligent filtering logic, data organization, and result presentation. Instead of manually scanning spreadsheets and cross-referencing criteria, analysts could now run the script and immediately see qualified opportunities.

Delivered 83% time reduction (30 minutes → 5 minutes), allowing the team to run more frequent analyses and catch time-sensitive opportunities. While smaller in scope than my other projects, it demonstrated the value of targeted automation for repetitive analytical tasks.