Yash Sachan

Building sub-millisecond order execution infrastructure for Nubra — an in-house fintech trading platform serving retail and institutional users.

Msg / sec

50K+

Msg / sec

P99 drop

35%

P99 drop

GC saved

40%

GC saved

Latency

<1ms

Latency

• 01Designed low-latency Golang services using advanced concurrency patterns — goroutines, channels, and worker pools — for high-frequency trading workloads.
• 02Built batch-based WebSocket pipelines that aggregate and stream market and trade data in real time, cutting network overhead and lifting client-side throughput.
• 03Used NATS JetStream for durable event streaming, and integrated Python alerting pipelines for live monitoring.
• 04Distributed the order book across custom ring-buffer structures, cutting GC pressure ~40% and P99 latency 35%, profiled with Go pprof.
• 05Integrated NSE, BSE, MF and IPO platforms over SOAP and Open APIs, enabling real-time data ingestion and order workflows.
• 06Engineered core Nubra modules — order management, portfolio tracking, real-time market feeds, plus eDIS, Early Pay-in and Pledge settlement flows.

• Golang
• NATS JetStream
• Kafka
• PostgreSQL
• Redis
• Python
• OpenTelemetry

Worked across the stack — Next.js frontend and high-performance Golang backend — on a fintech platform under production load.

Throughput

25%

Throughput

Manual work ↓

21%

Manual work ↓

Obs. tools

3

Obs. tools

• 01Contributed to the frontend architecture using Next.js and Material UI.
• 02Engineered high-performance Golang backend services using goroutines and channels for concurrent request handling — improving throughput 25% and reducing latency.
• 03Developed an internal chatbot with Go and Azure AI Services, using concurrency patterns for parallel workflow execution across multiple users, cutting manual tasks 21%.
• 04Enhanced observability with Grafana, Prometheus and Loki — identified slow endpoints and optimised hot paths.

• Golang
• Next.js
• Material UI
• Azure AI
• Grafana
• Prometheus
• Loki

Modernised a billing management platform — front to back — for faster loads and a more robust data layer.

Faster loads

30%

Faster loads

Next.js

SSR

Next.js

Deployed

GCP

Deployed

• 01Migrated the billing management system from React to Next.js, reducing page load times 30%.
• 02Designed and implemented a backend architecture using PostgreSQL, FeatherJS and GCP.
• 03Built typed, responsive UI with TypeScript and Tailwind CSS across the billing dashboard.

• Next.js
• React
• TypeScript
• Tailwind CSS
• FeatherJS
• PostgreSQL
• GCP