Hariom-Sahu

I build backend systems that stay fast when the load doesn't stay small.

Software Engineer at Multi Commodity Exchange. I design distributed data pipelines, streaming systems, and production GenAI applications.

How I work

Hariom Sahu

I turn fragile monoliths into resilient, observable pipelines. Most of my work sits where throughput, correctness and compliance meet: financial data that has to be right, fast, and auditable.

  1. Design for the failure mode

    Traffic spikes, partial outages and bad input will happen. Guardrails and observability ship with the first version, not after the first incident.

  2. Measure, then optimise

    Query plans, latency percentiles and profiler output decide the redesign. The 92.5% and 50% wins came from data, not guesses.

  3. Ship systems people can trust

    From JWT-gated gateways to LLM guardrails in a regulated environment, reliability and safety are requirements, not features.

Education

B.E., Computer Engineering

A. P. Shah Institute of Technology, University of Mumbai

2021 — 2025

  1. 2021Started
  2. 2022
  3. 2023
  4. 2024
  5. 2025Graduated

Toolbox

Languages

  • Python
  • Java
  • TypeScript
  • JavaScript

Frameworks & libraries

  • FastAPI
  • Node.js
  • React
  • SQLAlchemy
  • Celery

Data & streaming

  • PostgreSQL
  • MongoDB
  • Redis
  • Kafka

Infrastructure

  • Docker
  • AWS
  • Nginx
  • CI/CD

Where I've been

Software Engineer

Multi Commodity Exchange · Mumbai

Jun 2025 — Present

Production REST APIs and distributed data pipelines for client lifecycle management on India's largest commodity exchange.

Key achievements

File validation time
92.5%faster · 40 min → 3 min
API search latency
50%lower via indexing & query redesign
RAG pipeline latency
40%lower with 5-way parallel retrieval
Users on shipped APIs
100M+on India's largest commodity exchange

What I worked on

  • Re-architected a legacy monolith into a distributed Python / Celery / Redis pipeline, cutting file validation from 40 to 3 minutes.
  • Built a real-time Kafka ETL pipeline and anomaly-detection engine that raises multi-channel alerts on suspicious trading behaviour.
  • Halved API search latency through query optimisation, schema redesign, B-Tree indexing and SQLAlchemy refactoring.
  • Designed a custom Nginx API gateway with JWT authentication and rate limiting to shield downstream services from traffic spikes.
  • Provisioned and hardened UAT and pre-prod environments: Redis, Nginx reverse proxies, end-to-end logging and observability.
  • Built internal automation tools adopted company-wide, removing manual operational workflows.
  • Python
  • Celery
  • Redis
  • Kafka
  • PostgreSQL
  • Nginx
  • AWS

Systems I've shipped

All projects