Backend / platform / data infrastructure
I build backend platforms, event-driven workers, query paths, recovery tooling, and provenance systems for production data products.
Selected backend achievements
2026
The same executive arrives from many sources under different identifiers with conflicting values. Canonical person records rebuild nightly across about 6,000 companies with weighted source precedence and append-only column-level provenance, so a disputed field traces to its sources and a bad source can be retracted.
2026
Scaled screener coverage about 10x to nearly 100 metrics, each filterable as a number, text match, or selection in free combination, so the predicate space outgrew any index. Moved analytical reads to DuckDB over S3-backed Parquet, vertically partitioned so financials and management export in parallel and load independently.
2026
No shared replication existed between source and target. Repointed writers first, bulk-restored, then closed the write gap with a bounded catch-up pass, trading a stale-read window on upserted rows for continuous availability.
2026
Avro schemas with a schema registry. Cron emits trigger events so scheduled and reactive work share a single path, retry, and DLQ, with idempotent producers, ordered shutdown, and correlation-ID tracing across workers.
2026
Plans compose from features as data rather than code, entitlements materialize from subscriptions, trials, and referral rewards through one grant path, and every provider identifier is abstracted so a second payment processor requires no schema change.
2026
Structured logging drains Redis streams to S3 as Parquet, with correlation IDs spanning API and worker processes. Built after undrained container logs filled the production host and took Redis down; off-host retention replaced ad-hoc rotation.
2026
Restored deleted production rows from a point-in-time Neon branch through PostgreSQL dblink, then added a pre-commit migration scanner for destructive SQL patterns.
2025-2026
Separated transformation logic for estimates, sales orders, and invoices from persistence, added PostgreSQL audit partitioning, and used Rust Decimal types with exhaustive enums across a Python/Rust boundary to prevent rounding drift.
2025-2026
Preserved incomplete checkout attempts, added gateway-initiation audit trails, and built polling-based wallet recovery with transactional row locking so automated sweeps and admin resolution could not lose updates.
Experience spine
Feb 2026 - Present
Backend and distributed systems engineer, remote. Financial-data infrastructure: identity resolution, DuckDB/Parquet analytical serving, a Kafka event backbone with Avro, production PostgreSQL on AWS RDS, billing and entitlements, and structured logging with correlation IDs across API and worker processes.
Dec 2025 - Jan 2026
Software engineer, contract, remote. Rust precision-arithmetic pipeline using Decimal types and exhaustive enums to prevent financial rounding drift across a Python/Rust microservice boundary.
Sep 2025 - Jan 2026
Full-stack developer, contract, hybrid in Kathmandu. Owned backend/frontend integration for news, events, subscriptions, authentication, and reporting on a multi-tenant platform, implemented cross-domain authentication and business reporting, and restored production stability for a platform serving thousands of users.
Apr 2025 - Nov 2025
Promoted from Associate in April 2025. Backend services in TypeScript, Express, gRPC, Kafka, and PostgreSQL for high-throughput request and worker flows, JWT/RBAC and external identity-provider integrations, and system-design and onboarding documentation.
Mar 2024 - Apr 2025
Modular TypeScript services, secure authentication flows, query optimization with indexing, caching, batching, and constraints, shared utilities, API design, and teammate mentoring.
2022 - 2024
Freelance delivery of a project bank information-management system with React, Node.js, and PostgreSQL (Sep 2023 - Jan 2024), backend scheduling and message-queue email flows at Brandbuilder Nepal (Nov - Dec 2023), and React, TypeScript, PostgreSQL, and build tooling at Pagevamp Nepal (Dec 2022 - Mar 2023).
Education
2017 - 2023
Bachelor of Engineering in Computer Engineering, Tribhuvan University.
Academic systems projects
Built a protein contact-map prediction pipeline with sequence cleaning, duplicate removal, feature generation, reproducible train/validation/test splits, and CNN experiments; also built a face-recognition pipeline with preprocessing, augmentation, CNN embeddings, L2 normalization, and batch-wise cosine-similarity verification.