Kush Vijapure

Built an MCP server linking Claude to the bank's internal financial database, a Streamlit dashboard tracking model pipelines across active deals, and a tool normalizing client exports into summaries — adopted across the deal team.

Improved benchmarking latency and accuracy, built reference solutions and test suites in Python/SQL, and reviewed AI-generated code for logic errors and anti-patterns.

Built Python tools automating pricing, client tracking, and reporting across ~$650K annual revenue, plus an LLM model on ~100 acquisition calls that raised close rates from 9% to 20%.

Built an MCP synced to Tabroom and OpenCaselist tracking opponent tendencies and judge biases, plus custom Verbatim macros — as head of pre-round strategy for five top-10 and two top-4 national placements.

Selected as 1 of ~30 for a competitive public-markets program, including a two-week NYC networking trip with top investment banks.

Produces monthly stock pitches, financial models, and decks for the club's $70K+ L/S equities fund, engaging 100+ members.

Tutored students and ran ML workshops — including coaching the student who won the A&M AI Pokémon bot competition.

Attended weekly sessions with legal professionals and Harvard Law workshops on admissions, the LSAT, and applications, and helped route prospective students into the pre-law advising pipeline.