Aniket
Shrivastav
Backend Systems ยท Distributed Architecture ยท AI-Powered Engineering
By day I build cross-border payments & FX infrastructure at Aspire. On the side I'm building Nideknil, an AI hiring platform that ranks every candidate for a role using vector retrieval, reranking and LLM evaluation.
Engineer. Problem Solver.
AI Enthusiast.
I'm a Software Engineer with deep expertise in distributed systems, microservices, and fintech backend architecture. Currently at Aspire, building cross-border payments and FX infrastructure across SG, HK, AU, US and SEA. Previously at WebEngage and Groww.
I graduated from Indian Institute of Information Technology Nagpur in Computer Science, and am passionate about competitive programming โ 4โ on CodeChef.
Outside work I'm building Nideknil ("LinkedIn" spelled backwards), an AI hiring platform I designed and shipped end to end. It covers a candidate-ranking funnel, a merit-ranked referral marketplace, and AI-fluency scoring that uses Claude Code and MCP.
Tools I wield.
Where I've made an impact.
Click any role to see the full impact story.
- Wallex SEA multi-region launch: built IDR/PHP/MYR/VND collections, FX conversion and local payouts end to end for HK, AU and US clients in Java and Golang. Includes per-region credential isolation, idempotent Kafka webhook processing and automatic rollback of failed legs. Launched in 3 new regions with FX into 13 currencies.
- HK migration engine: a resumable, step-tracked Java workflow that moved 1,800+ clients from CC-SG to CC-HK. It automates balance transfers, account creation, recipient migration and ledger sync, with safe re-runs and retries that respect rate limits. Cut manual ops by 60%.
- Concurrency-safe transaction limits: configurable outbound limits for POB-unverified businesses across every transfer flow (single, bulk, scheduled, bill and claim). Distributed locking prevents concurrent overspend, with maker-checker controls and rollback on failure. Boosted new-business activation by 42%.
- Multi-rail payout service: DBS-SG ACT, DBS-HK TT and CC-HK transfer flows in Golang with provider-level error handling and retries. Reduced cross-border FX transfer failures by ~30%.
- Huifu CNY payment rail: a new cross-border CNY corridor built end to end, covering recipient creation, transfer dispatch, fee calculation and Unified Social Credit Code validation. Opened China payouts for HK clients.
- Recipient platform at scale: rebuilt recipient create/update/delete/migrate flows with a zero-downtime MySQL migration, query optimization and caching on
GET /v1/counterparties. Added fallback provider routing for when Payout Hub is disabled. Cut p95 latency by ~70% and scaled to 30,000+ businesses. - Observability & reliability: integrated Sentry, OpenTelemetry and Datadog across Aspire Hub, and built AWS S3 statement ingestion with exponential-backoff retries and Slack alerts. Cut incident MTTR by ~50%.
- Built ConvertWise Backend APIs in Java/Spring โ AI-based personalized campaigns that increased monthly sales by 25%.
- Developed Wise Cart โ AI-enabled cart boosting upsell and personalized reward banners โ enhanced 15% AOV for Shopify merchants.
- Re-architected Historical Data Sync (HDS) service using Dropwizard & AWS โ reduced data-stuck problems by 50%.
- Led Zid Integration Service โ full DB schema design and code from scratch to power recommendations for Zid App merchants.
- Built GrowwPay UPI service โ send/receive money flows with collect request APIs in Java & Spring Boot.
- Developed event publisher for push & in-app notifications via Kafka producers/consumers.
- Built webhook manager service to receive and process payloads using gRPC and Apache Kafka.
Nideknil: LinkedIn, flipped.
An AI hiring platform I designed, built and shipped solo. Recruiters don't hunt through profiles; AI ranks the whole candidate base for any role.
Screening every applicant with an LLM is slow and expensive, so Nideknil treats the LLM as the scarce resource. Cheap stages cut the pool down first, and only the top handful of candidates reach deep AI evaluation. Every score comes back explainable, factor by factor.
Wins beyond work.
Ready to work together?
Whether it's a full-time role, an AI project, or a freelance sprint โ I'd love to hear what you're building.