# Bijay Subedi > Senior Rails & React engineer in Kathmandu, Nepal., Kathmandu, Nepal (UTC+5:45). Building AI-powered products and smarter engineering workflows, and turning operations run on spreadsheets into software teams rely on. - Availability: Open to senior full-time remote roles; overlap 4+ hours with US Eastern · full EU workday - Email: sharma.bj11@gmail.com - Résumé: https://bijay3030.github.io/resume.pdf - GitHub: https://github.com/bijay3030 - LinkedIn: https://www.linkedin.com/in/bijay-p-subedi/ ## About I joined Truemark in 2021 as an associate engineer and was leading client applications as a senior 16 months later. Most of that work serves US healthcare clients, where one lost file or wrong price breaks trust. I start by mapping how the work actually moves: who touches what, and where it breaks. Then I model it as explicit states and automate every step between. ### How I ship a feature with AI 1. Understand: the requirement, and the problem behind it 2. Research: how others have built it 3. Plan with AI: frontier models draft the plan in plan mode 4. Grill the plan: Q&A rounds where the model asks every question before any code 5. Slice: the plan becomes small, verifiable tasks 6. Agents implement: one task at a time 7. QA: I verify the work myself 8. Close the gaps: anything missed goes back to the agent until it’s right ## Selected work ### Helios — Medical translation One platform for medical translation projects, from file intake to invoice. - Before → after: Email threads and spreadsheets → One auditable workflow - Scale: ~20 files per project; 4–5 language pairs; up to ~100 tracked tasks - Client: US healthcare localization provider (NDA) - Problem: Every file, in every language, needed its own vendor, deadline, and delivery. Coordinating that by hand was slow, risked lost files, and left no audit trail for clients who require one. - My part: Sidekiq pipeline that validates and routes multilingual files on upload - My part: Automated pricing, turnaround, and vendor work orders - My part: Live dashboards and role-based access - Stack: Rails 7, React, Sidekiq, ActionCable, AWS S3, Pundit, Docker - Decision: Background jobs for anything slow. Uploads return instantly; failed validation or routing retries safely. - Decision: Authorization as policies. One Pundit policy per model decides who can do what. Essential when clients, vendors, and staff share one system. - Decision: Push, don’t poll. Dashboards update the moment status changes, with no polling load. - Result: Every hand-off automated. - Result: Audit-ready history for every job. - Result: More volume without more coordinators. ### Quoting — Localization pricing Turns a translation request into an accurate, versioned quote. - Before → after: A spreadsheet per project manager → One shared rulebook - Scale: larger file sets than Helios; large worksheets per project type - Client: Same client as Helios - Problem: Price depends on word counts, language pairs, schedule, and client-specific rules. Each PM priced from a personal spreadsheet: quotes were slow, one job could get two prices, and a change request had no reliable record of the previous quote. - My part: Pricing-rule and version data model in PostgreSQL - My part: Multilingual line-item editor with live co-editing - My part: Integrations: external intake in, delivery out - Stack: Rails 7, React 18, PostgreSQL, ActionCable, AWS S3, Pundit, Docker - Decision: Immutable versions, not edits in place. History, comparisons, and client disputes become simple queries. - Decision: Pricing rules as data, not code. A new client or project type is a data change, not a deploy. - Decision: Real-time only where it matters. Live sync runs on the editor alone, where several people touch one quote. - Result: Faster quotes, less back-and-forth between PMs and sales. - Result: One price per job, whoever quotes it. - Result: Full revision history for enterprise clients. ### aListEngine — AI cataloging Photos in, auction-ready listings out, for auctioneers and resellers. - Before → after: Hours of writing per batch → Minutes, reviewed by a person - Scale: live product; hundreds of photos per batch; 10 export platforms; iOS + Android - Client: Auctioneers, estate-sale teams, resellers — https://alistengine.com/ - Problem: Every one-off item (antiques, collectibles, equipment) needs a title, description, condition notes, and price, formatted differently for each platform. By hand, quality depended on the writer. - My part: Photo-to-draft AI pipeline - My part: Rules engine for seller fields and marketplace formats - My part: Shopify, AuctionFlex, and LiveAuctioneers exports - Stack: Rails, React, Docker - Decision: The AI drafts, a person approves. Quality stays high; mistakes stay cheap. - Decision: Marketplace rules outside the prompt. Code enforces formats, so a new marketplace never means rewriting prompts. - Decision: One adapter per marketplace. Adding a channel is isolated work. - Result: Consistent quality across sellers and channels. - Result: Paid product with a free tier and 14-day trials. ## Experience ### Senior Software Engineer, Truemark (Aug 2022 to now) Own each application end to end, from data model to deploy pipeline. - 6+ client Rails applications led - 99.9% API uptime at 500K+ requests a month - −40% inter-service latency, monolith → AWS microservices - <8 min deploys, down from ~45 after a CI/CD rebuild - −60% critical query time, PostgreSQL schema and index redesign - 87% test coverage (RSpec, Jest); bug reports down 35% - Introduced Hotwire and tuned React: initial load ~35% faster, no full-page reloads in key flows. - Mentored 3 junior engineers through weekly Rails and React reviews. ### Associate Software Engineer, Truemark (Apr 2021 to Aug 2022) Backend foundations for high-traffic client apps. - Shipped Rails APIs and services with consistent sub-200ms responses. - Integrated 10+ third-party services: payments, analytics, communication. - Improved React frontends: client-reported UI issues down 25%. - Helped the team adopt Scrum; kept technical documentation current. ## Education - B.Tech, Computer Science, Maharshi Dayanand University (2016 to 2020) ## Toolkit - Backend: Ruby on Rails 7–8, PostgreSQL, Redis, Sidekiq, ActionCable, REST APIs - Frontend: React 18–19, TypeScript, Hotwire (Turbo + Stimulus) - Infrastructure: AWS (EC2, S3, Lambda, RDS), Docker, GitLab CI, GitHub Actions - AI: Agentic coding, LLM fundamentals, Production AI pipelines - Quality: RSpec, Jest, TDD ## Frequently asked questions ### Who is Bijay Subedi? Bijay Subedi is a senior software engineer based in Kathmandu, Nepal, with 5+ years of experience building Ruby on Rails, React, and AWS applications. Bijay works at Truemark, building workflow and automation platforms for US clients. ### What does Bijay Subedi specialize in? Full-stack Ruby on Rails 7 and React applications: Sidekiq background processing, real-time updates with ActionCable and Hotwire, PostgreSQL performance tuning, AWS microservices, and CI/CD pipelines. Bijay builds with AI daily: he researches each feature, plans it with frontier models in plan mode, resolves open questions through Q&A rounds, splits the plan into small tasks for AI agents to implement, then verifies the work himself and sends any gaps back to the agent. ### What has Bijay Subedi built? Helios, a platform that runs medical translation projects from intake to delivery; Quoting, a versioned quote system for a healthcare localization company; and aListEngine (alistengine.com), AI cataloging software that turns item photos into auction-ready listings. ### Does Bijay Subedi work remotely with US or European teams? Yes. Bijay already works remotely with US clients through Truemark and is based in Kathmandu, Nepal (UTC+5:45), with 4+ hours of overlap with US Eastern and a full EU workday. Bijay is open to senior full-time remote roles. ### How can I contact Bijay Subedi? Email sharma.bj11@gmail.com.