Sulav Raj Bista

Remote

• Served as technical owner of the Vendor Management module — a core domain spanning authentication, onboarding, tax compliance, and payments — and the single point of escalation across engineering, product, support, and fraud teams.
• Architected a 2FA authentication framework using Auth0, Twilio, and Google Authenticator that mitigated vendor account fraud and saved an estimated $1M+ annually in fraud losses.
• Sustained a 99.99% API success rate and p95 latency under 700ms across the module under peak production load, owning module-wide SLOs end to end.
• Designed asynchronous tax-validation microservices for US and international markets via third-party APIs and webhooks, reducing compliance penalties by an estimated $100K+ annually.
• Built a self-serve vendor onboarding flow with Django and OneSchema, cutting onboarding time by 86% and eliminating recurring manual toil.
• Implemented Redis-backed distributed caching that improved API throughput by 45% and absorbed peak traffic without scaling infrastructure spend.
• Built internal tooling adopted across Engineering, Support, and Fraud — vendor impersonation, bulk onboarding, and a 2FA-reset workflow — eliminating dozens of weekly engineering interrupts.
• Grew test coverage by 54% through Pytest suites and legacy refactoring, and established observability across Datadog, Metabase, and Sentry that reduced critical-incident MTTR by 40%.

Python, Django REST Framework, PostgreSQL, Celery, Redis, Kafka, ElasticSearch, Heroku, React, Auth0, Datadog, Sentry, Pytest

Lalitpur, Nepal

• Built and scaled a high-traffic Flask backend on AWS for dynamic ad-publishing portals, handling peak concurrent request loads across multiple client portals.
• Developed real-time analytics dashboards that improved client revenue tracking and supported data-driven decision-making.
• Designed a service-oriented architecture with optimised inter-service communication, improving system modularity and fault tolerance.

Python, Flask, PostgreSQL, Node.js, React, Prisma ORM, AWS

Bangalore, India

• Improved OCR receipt-classification accuracy from 35% to 95% through model tuning and pipeline redesign, enhancing UX and reducing manual processing overhead.
• Designed and launched a secure ACH payment service integrating third-party APIs and webhooks, with a micro-deposit payee-verification framework.
• Engineered automated statement parsers for Visa and Amex transactions delivered over SFTP, processing thousands of financial records reliably.
• Scaled high-traffic REST APIs through targeted database indexing and query optimisation, reducing average response times by 40%.

Python, Flask, Java Dropwizard, PostgreSQL, AngularJS, Docker, Kubernetes, RabbitMQ, AWS

Bangalore, India

• Delivered client-facing features and resolved production issues on a legacy financial platform, collaborating closely with QA and product teams within Agile sprint cycles.

C#, ASP.NET Framework, MS SQL Server, LINQ

• Built a real-time collaborative PWA (Next.js, React, Supabase) with optimistic updates and an offline IndexedDB write queue that reconciles via last-write-wins on reconnect; in daily use by several households.
• Engineered timezone-correct recurring reminders with pg_cron and an RRULE subset in Postgres, plus a three-tier item categoriser (keyword dictionary → rate-limited, cached Gemini LLM fallback).

Next.js, React, TypeScript, Supabase (PostgreSQL/Realtime/Auth/Edge Functions), Tailwind, Vercel Case study →

• Independently designed, built, and launched a production SaaS for a private tax-collection firm: Django REST Framework with JWT-based RBAC and cursor-based pagination, plus a React frontend with client-side image compression before presigned S3 upload.
• Re-platformed from AWS (EC2/Docker/RDS/Amplify) to a lean self-hosted stack (Hetzner + systemd/Gunicorn, self-hosted PostgreSQL, Cloudflare), cutting monthly running cost to a fraction.

Django REST Framework, React, PostgreSQL, AWS, Hetzner, Cloudflare Case study →

• Shipped 9 client-side tools with an off-main-thread compression pipeline (Web Workers + OffscreenCanvas, zero-copy ImageBitmap transfer) and in-browser PDF workflows (pdfjs-dist, pdf-lib) for files up to 100 MB — no uploads, no backend.

Next.js, TypeScript, Web Workers, pdf-lib, Cloudflare Pages Case study →

Ramaiah Institute of Technology · Bangalore, India

• GPA 8.73/10. Awarded a full Merit Scholarship by the Government of India.
• Published research on "Predicting the Priority of Bug Reports Using Classification Algorithms" in the IJCSE journal.