apoorva kumar

Built a talent qualitative metrics system and multi-stage talent filtration pipeline, integrating candidate signals into structured data and transparent shortlist reasoning. Created a learning layer trained on real career-counsellor session data to improve filtration accuracy over time.

Designed and built an SLM-powered research synthesis engine that extracts structured insights linked to supporting interview quotes, reducing analyst effort. Implemented full observability with event tracking and health metrics, and used it as a feedback loop to automate post-research analytics, saving 12+ hours.

Led user research and built automated pre/post-session communication flows across WhatsApp and email, personalized by GA/Mixpanel/CX-derived segments and session type. Developed retention analytics and KPI dashboards in Metabase/Mixpanel, driving a 30% increase in repeat orders and CSAT improvement.

Engineered a high-throughput tick-by-tick ingestion pipeline in Golang using Apache Flink and TimescaleDB, including columnar storage with Apache Parquet and low-latency windowed aggregations. Built a centralized data platform for trading back-tests and optimized query performance, delivering 20% reduction in net trading losses and saving 30+ lakhs/week.

Led migration of Ratings & Reviews from a PHP monolith to Golang microservices with MySQL to MongoDB and batch jobs to Kafka queues, improving review frequency by 70% and rating accuracy by 85%. Built critical API segregation with multi-layer caching for booking, delivering 100K+ concurrent users at ~3ms response time and improving the booking funnel by ~40% while protecting ~₹30L in peak-event sa

Built product modules for Loannxt (B2C credit platform) using PHP and Node.js, and supported technical operations. Managed security protocols and process automation to improve delivery reliability and operational hygiene.

Grade: 7.6

Earned a B.Tech from Maharishi Markandeshwar University, graduating in 2017 with a CGPA of 7.6.