Devatraj Bhattacharjee

What I do, 4 of 5

Backends that hold up

Event-driven backends that keep working when traffic spikes or a server dies.

10M+requests a day on a platform I built

Who it's for: Engineering leaders hiring for backend and platform work.

The problem

Try it

A small working tool. Change the numbers.

Break it and watch it recover

Events flow through a topic with three partitions to three consumers. Spike the traffic or stop a consumer. Nothing gets lost: the group rebalances and works through the backlog.

0events processed
0waiting (consumer lag)
0events lost

All 3 consumers healthy.

Where I've done it

Jeevansathi

Firebase Remote Config, replaced

Built an in-house feature-configuration platform (Spring Boot, Redis, MongoDB) that replaced Firebase Remote Config after it moved to a paid tier.

  • $70,000/year licensing cost eliminated
  • 10M+ requests/day

Jeevansathi

Event-driven matching on Kafka

Designed a fault-tolerant, event-driven matching algorithm on Kafka and NoSQL stores.

  • 500,000+ users connected

Jeevansathi

Real-time GenAI matchmaking

Delivered a real-time GenAI matchmaking engine using Spring Boot, Openfire (XMPP) and Aerospike, orchestrating GPT, Gemini and Qwen through a LangChain/LlamaIndex RAG pipeline on RedisVL vector search.

  • 99.9% uptime

99acres

API aggregator

Standardised cross-service calls through a containerized Spring Boot API aggregator with unified headers.

  • 15% lower API latency

Jeevansathi

Breadth across production services

Shipped changes across dozens of production microservices.

Config panelIllustration
home_screen_variantAndroid
  1. app version = 33.5"off"a build with a known bug
  2. app version >= 34.0"v2"new home screen
  3. app version >= 33.0"v1"
  4. no rule matchesapp default

The app gets "v2"

    Config platform. First matching rule wins, served from Redis, 304 when nothing changed.

    How I'd do it for you

    1. Decouple with events

      Services talk through Kafka or RabbitMQ, so one slow consumer doesn't block the rest.

    2. Cache what's hot

      Redis and ETags keep repeat reads off the database.

    3. Fail safe

      Every dependency has a timeout and a fallback.

    Need this?

    Tell me what you're working with. I'll tell you honestly whether it's worth doing and roughly what it takes.

    Email me about this