Open to full-time rolesGurugram, India · remote-friendly
~/ snehil-sharma — backend engineer

I build backends, and the interfaces that sit on top of them.

Real-time video infrastructure at LenscorpAI — Go microservices behind 15,000+ CCTV cameras across enterprise deployments on three continents, held at 99% uptime. Before that, backend performance and tooling at Stockarea.

Résumé
snehil@portfolio: ~
$
live · deployments
18.0° N   40.0° E
15+ sites · 3 continents
in production
15,000+
CCTV cameras, 99% uptime
now
SDE · LenscorpAI
Leading a 4-engineer backend pod
0+
Years experience
0
Shipped projects
0
Production companies
01 / experience

Where I've worked

Two companies, one obsession: software that holds up under real production load.

Jan 2025 to present

Software Development Engineer

LenscorpAI

Backend for an AI-powered surveillance platform: Go microservices running CCTV and inference infrastructure across enterprise deployments on three continents.

  • Architected and shipped 10 production Go microservices (AWS EC2, S3, Docker) across 15+ enterprise deployments in India, the US, Saudi Arabia and Egypt, managing 15,000+ CCTV cameras at 99% uptime on air-gapped, offline-first infrastructure
  • Built a real-time video ingestion layer (WebSockets, Redis, Go) streaming 8,000-10,000+ camera feeds into a GPU-offloaded AI inference pipeline (face recognition, intrusion detection, line crossing, GIS), cutting system CPU usage from 85% to 58% with concurrent Go services
  • Built a Kafka-based messaging pipeline for real-time AI alerts and license-update distribution across all 15+ sites, with at-least-once delivery
  • Engineered an S3-backed remote update system for the Electron desktop app, delivering live config and binary updates across 15+ globally distributed field deployments
  • Lead a 4-engineer backend pod as primary technical decision-maker: code review, sprint planning and end-to-end delivery
GoTypeScriptElectronAWSDockerRedisKafkaWebSocketsNginxTunneling
Jan 2024 to Dec 2024

Software Development Engineer

Stockarea

Backend performance and tooling for a digital warehousing and logistics platform.

  • Diagnosed and eliminated memory leaks and N+1 query bottlenecks under sustained enterprise load (AWS RDS, Docker), cutting p99 query latency from 3s to 0.5s, an 83% reduction
  • Redesigned backend search logic and indexing strategy, reducing search latency by 68% across the platform's core discovery flow
  • Built a RBAC-enabled Excel reporting tool (Docker, AWS RDS) serving 20+ admin and finance users with role-filtered exports and zero manual intervention from engineering
  • Integrated Zoho Books, E-way Bill and Toll APIs to automate accounting and logistics workflows, improving operational efficiency by 30% and tracking accuracy by 95%
MySQLTypeScriptRedisAWS RDSAWS EC2Docker

Want the full picture? Grab the résumé: one page, no fluff.

Download résumé
02 / projects

Things I've built

Four projects, from retail demand forecasting to production streaming infrastructure. Hover or tap one for the case study; arrow keys work too.

Full-stack / Data · 2026

ShelfDekho

Full-stack engineer

ShelfDekho screenshot

A retail intelligence dashboard that tells a grocery chain what will run out before the shelf is empty — demand forecasting, stockout alerts and reorder quantities for every product in every store.

the problem

A chain loses sales when a shelf empties before the next delivery, and a plain revenue figure hides why: a festival that shifted dates or a single stocked-out SKU looks the same as a real decline.

the approach

An eight-week, calendar-aware demand forecast per SKU per store (Diwali moves every year), paired with stockout prediction from sell-through rate versus supplier lead time, and a natural-language layer that answers 'why did sales move?' with the numbers behind it.

what I built
  • Per-SKU, per-store demand forecast that beats naive baselines (17.7% vs 19.8-26.3% error)
  • Stockout prediction: days-to-empty vs lead time, surfaced as a live shelf grid
  • Reorder engine with configurable safety stock and supplier case sizes
  • Revenue attribution that separates festive and promo effects from real movement
  • Natural-language 'ask' over the data, answered with charts and figures
Next.jsTypeScriptPythonForecastingLLM
Live site
03 / stack

What I work with

Languages

Core languages

GoJavaPythonTypeScriptSQL

Backend & Systems

Server-side & APIs

Node.jsSpring BootMicroservicesDistributed SystemsWebSockets

Databases & Messaging

Data & streaming

PostgreSQLMongoDBRedisElasticsearchKafka

Cloud & DevOps

Infrastructure

AWS (EC2, S3, RDS)DockerCI/CD

AI / ML

Inference & models

TensorFlowHuggingFaceONNX RuntimeVLMsReal-time inference
04 / contact

Let's build something

Open to full-time roles and select freelance projects. Email is the fastest way to reach me.