Hey, I'm Yashvi

I'm a backend software engineer based in Bangalore, India, with 3.5 years of experience building systems that scale. Currently at Abnormal AI , previously at Gojek . I studied Computer Science at IIT Jodhpur.

Outside of work, I'm drawn to psychology and philosophy — understanding why people think and act the way they do. I love reading, meeting new people, and picking up offline games like cards and badminton. I want to create impact at scale and help build things that matter.

Work Experience
See all work
  • Sep2024 - Current
    Abnormal AI
    Software Engineer

    Founding engineer for AI Phishing Coach (AIPC) — built 0 to 1 for an LLM-powered phishing simulation product that trains enterprise employees against phishing, BEC, and credential theft.

    Owned and built the entire engagement tracking system from scratch in Go — a public-facing service tracking email opens, link clicks, and credential submissions. This became the Go service boilerplate adopted org-wide.

    Optimized report generation by 168x (56 min → 20 sec for 40k records) by redesigning sequential I/O into concurrent batch processing. Built the Simulation Grouping feature that revamped the product’s settings module and was critical for closing enterprise deals.

    Designed an image archival pipeline spanning Airflow DAGs, a concurrent Go cron job, and S3 + CloudFront CDN to fix broken template images across all tenants daily.

    Tech: Django, Python, Go, Kafka, Airflow, PostgreSQL, gRPC, AWS, Prometheus, Grafana

  • Jul2022 - Sep2024
    Gojek
    Software Engineer

    Developed a Ruby-based annotation system to enhance Google reCAPTCHA accuracy, reducing false positive rates from 12% to 1.8% and improving DDoS attack mitigation.

    Refactored Go microservices and optimized infrastructure for cost-saving initiatives, saving $385 monthly. Built features to improve bid receive rates for 20% of drivers processing 1.5 million records daily.

    Developed a feedback processing service in Go — the most requested feature by drivers — to remove unrelated feedbacks and improve fairness. Improved driver payout deduction reliability from 93% to 100%.

    Tech: Go, Ruby on Rails, TypeScript, Redis, GCP, Kafka, React

Let's Connect

If you want to chat about engineering, philosophy, or anything interesting — reach out on social media or send me an email.