ratneshHiring? I build backends that still hold at 2 a.m.

नमस्ते, I’m

Ratnesh Maurya

Backend & AI engineer at Initializ · Gurugram, India

NOW
SDE at Initializ — core backend engineer on the Nestara lending platform and its CRM
BUILDS
Distributed backends · AI agents & orchestration · Kubernetes operators
OPEN SOURCE
Forge — WhatsApp channel for the agent runtime (PR #452)
BASED IN
Gurugram, India · IST (UTC+5:30)

I build the backends lending and enterprise platforms run on — Go and Elixir services, Postgres, Kafka, Kubernetes — and the AI agents on top that do real work, not just chat.

Email me →Résumé
GitHub ↗LinkedIn ↗X ↗Blog ↗
shipping production backends
3+ yrs
commits across Initializ & Nestara repos
2,200+
commits to the Nestara lending backend — #1 committer
1,436
line open-source PR to Forge
6.2k
§ GITHUB · @ratnesh-maurya · LAST 12 MONTHS

2,494 contributions in the last year

active days
303
longest streak
20d
current streak
6d
best day
34
Hover a day to see its count.Less Moregithub.com/ratnesh-maurya ↗
§ 01 — WHAT I DO

Backends, and the agents on top.

Three things I get hired for. Most of my work lives in the first two at once.

Backend & distributed systems

Loan journeys, credit-bureau risk engines, lead routing and event-driven notification pipelines — services that move money-adjacent data and cannot drop a message.

Go · Elixir / Phoenix · PostgreSQL · Kafka · Redis · Oban

AI agents & orchestration

Agents that call real tools: MCP servers, hand-written tool-calling loops, typed pydantic-ai agents where the model never does the arithmetic, A2A agents and RAG over Qdrant.

MCP · pydantic-ai · OpenAI · A2A · RAG · Qdrant

Platform & reliability

Kubernetes operators, secret management, and the observability that finds the leak before users do.

Kubernetes · AWS · Docker · pprof · Prometheus · Grafana
§ 02 — SELECTED WORK

Things running in production.

Client and company work first; side projects below. Most client repos are private — links go to what is public.

LEAD STORY · Client platform · LendingSep 2024 — present

Nestara — the lending backend, its CRM, and the AI on top

Nestara helps home buyers get loans from many lenders. I’m the top committer on the backend that runs it and I created the CRM its sales team works in every day.

  1. i.Loan journeys, lender offers and DSA routing in Elixir/Phoenix on Postgres, with Oban and Redis job queues.
  2. ii.A unified CIBIL + Experian bureau report and risk verdict.
  3. iii.Nestara OneCRM: lead workspace, calling workflow, analytics.
  4. iv.An MCP tool server and a typed pydantic-ai agent for loan advice.
onecrm.nestara.in ↗nestara.in ↗
Elixir · Phoenix · PostgreSQL · Oban · Next.js · Python
backend commits, #1 of the team
1,436
journey events in the notification pipeline
19
credit bureaus merged into one verdict
2
repos: API, CRM, panel, web, app
5
AI agent 2026

Ask Nestara AI — an agent that never does maths

A pydantic-ai agent that must call a typed calculator command before it can answer. Exact-Decimal EMI, eligibility and balance-transfer engines with reverse solvers; rate limits, token budgets and a circuit breaker around the model.

Python · pydantic-ai · FastAPI · Redis · Postgresnestara.in ↗
Open source · initializ/forge 2026

Forge — WhatsApp channel for the agent runtime

Added WhatsApp as Forge’s fourth channel, in pure Go: QR pairing state machine, closed-by-default admission, a dedup ring as the reply-loop guard, and 16 end-to-end pipeline tests. Also fixed tool entrypoint resolution in the runtime.

Go · whatsmeow · Bubble Tea · AgentsPR #452 ↗
AI orchestration 2025

Loan-advisor MCP server + agent client

A FastMCP server exposing lender-offer, comparison, EMI and email-drafting tools, and a hand-written OpenAI tool-calling loop over MCP Streamable HTTP with one session per lead.

Python · FastMCP · OpenAI · DockerPrivate
AI agent · Initializ 2026

LogsRCAAgent — root cause from robot logs

An A2A-protocol agent in Go that finds why warehouse robots stopped: concurrent multi-pass log parsing, token-budgeted LLM batches, findings in Qdrant and MongoDB. RCA turnaround down 60%.

Go · A2A · LLM · Qdrant · S3Private

Side projects

All projects and open source →
Personal Tracker screenshotPersonal Tracker ↗tracker.ratnesh-maurya.com

Habits, expenses, sleep and study in one PWA with its own analytics.

JSONic screenshotJSONic ↗jsonic.ratnesh-maurya.com

Fast JSON formatting, validation and diffing on real payloads.

Rehabify screenshotRehabify ↗rehabify.ratnesh-maurya.com

Patient records and treatment tracking for de-addiction centres.

§ 03 — EXPERIENCE

Where I’ve worked.

B.Tech in Computer Science, AKTU (2020 — 2024), 8.0 CGPA.

Jul 2024 — PresentGurugram, India

Initializ

Software Development Engineer

Initializ builds AI-agent infrastructure. I split my time between Initializ’s own agent products and the Nestara lending platform, Initializ’s client.

Nestara (Loan Network Technologies) — lending platform, CRM & AI

Sep 2024 — present · 1,700+ commits
  • #1 committer on the Elixir/Phoenix lending backend (1,436 commits): loan-application journeys, lender offers, DSA lead routing, Hot/Warm/Cold lead priority, and an Oban-driven notification pipeline fanning out to WhatsApp, SMS, email and push.
  • Bureau risk engine — merges CIBIL and Experian into one report and one verdict across DPD, credit events and enquiries.
  • Created Nestara OneCRM (onecrm.nestara.in), the sales team’s CRM: lead workspace, calling workflow, credit cards and source-performance analytics, with encrypted API responses.
  • Ask Nestara AI — a pydantic-ai agent forced to call typed EMI, eligibility and balance-transfer tools before it answers, so the model never does the arithmetic; rate limits, token budgets and a circuit breaker.
  • Loan-advisor MCP server (FastMCP) with offer-comparison, EMI and email tools, plus a hand-written OpenAI tool-calling client keeping one MCP session per lead.

Initializ platform & agents

2024 — present
  • LogsRCAAgent — an A2A agent in Go doing root-cause analysis on warehouse-robot logs for GreyOrange: concurrent multi-pass parsing, token-budgeted LLM batches, results in Qdrant and MongoDB; cut RCA turnaround by 60%.
  • Built a notification microservice on Kafka, Redis and Protobuf with idempotent consumers — 99.87% exactly-once delivery.
  • RAG ingestion for Initializ Assistants (presigned S3 uploads, SharePoint/OneDrive connectors) and a fully self-hosted RAG + LLM chat platform on Ollama, Qdrant and MinIO.
  • Profiled Go services with pprof, Prometheus and Grafana; found leaks and cut heap allocation by 30%.
Go · Elixir · Python · PostgreSQL · Kafka · Redis · Qdrant · Kubernetes · AWS
Aug 2023 — Jun 2024Gurugram, India

Initializ

Software Developer Intern

  • Built the Kubernetes secrets operator from scratch (CRDs, open source) and the secrets API behind it; update latency down 67%.
  • Client-side encryption in TypeScript with the Web Crypto API for zero-knowledge secret storage.
  • Top contributor to docs.initializ.ai (146 commits) — Docusaurus, Algolia search, SEO.
Go · Kubernetes · TypeScript · Docusaurus
Mar 2023 — Aug 2023Remote

EMSEC

Software Developer Intern

  • Flask REST APIs and an SSL-certificate scraping pipeline across 300+ domains.
Python · Flask
§ 04 — WRITING · RATN LABS

Notes from the backend.

I write about system design, Go and what I learn shipping — at Ratn Labs.

All posts →blog.ratnesh-maurya.com ↗
§ 05 — CONTACT

Let’s build something.

Hiring for backend or AI-platform work, or need a hand with an agent that has to be right? Email is fastest — I reply within a day or two.

hiratneshmaurya@gmail.com ↗
Email meRésumé