The same one page as the PDF. Printing this page gives a plain copy without the border.

Ratnesh Maurya

Backend & AI Engineer (Golang, Elixir, Distributed Systems, AI Agents)

Technical skills

Languages
Golang, Elixir, Python, TypeScript, SQL, C++
AI & Agents
AI Agents, MCP (Model Context Protocol), pydantic-ai, OpenAI Tool Calling, A2A, RAG
Backend & Systems
Microservices, Phoenix, Kafka, gRPC, Protobuf, REST APIs, Oban, Concurrency
Databases
PostgreSQL, Redis, Qdrant (Vector Search), MongoDB
Infrastructure & Obs
AWS, Kubernetes (Operators/CRDs), Docker, Prometheus, Grafana, pprof

Experience

Initializ Pvt. Ltd. | Software Development Engineer

Gurugram, India | July 2024 – Present
  • Client – Nestara (home loans): top contributor (1,400+ commits) to the Elixir/Phoenix loan backend.
  • Built the credit-risk engine merging CIBIL and Experian into one verdict, so sales calls the best leads first.
  • Automated lead routing to sales agents and a notification pipeline (WhatsApp, SMS, email, push) for 19 stages.
  • Created Nestara OneCRM from scratch – the CRM the sales team uses daily for leads, calls and source analytics.
  • Built a 3-pass top-up loan pipeline (valuation, address extraction, geocoding, clustering, lender matching) that uses an LLM to clean and classify raw bureau addresses in concurrent batches, with a fallback when the model is down.
  • Secured the CRM APIs with AES-256-GCM response encryption, lead-level authorization and an audit trail; integrated Surepass CIBIL (cached), ParkPlus, Zapkey, Amazon SES and FCM.
  • Built “Ask Nestara AI”, a home-loan AI agent (pydantic-ai, FastAPI) that must call exact EMI and eligibility calculators before answering, so the LLM never guesses numbers; added Redis rate limits, token budgets and a circuit breaker.
  • Designed AI agent orchestration: an MCP server of loan-advisor tools (compare offers, EMI, draft email) and a custom tool-calling agent loop with one session per customer.
  • Built LogsRCAAgent, an AI agent in Go that reads GreyOrange warehouse-robot logs and explains failures; concurrent goroutine parsing cut root-cause analysis time by 60%.
  • Built a Kafka/Redis/Protobuf Notification Microservice with 99.87% exactly-once delivery; profiled Go services with pprof, Prometheus and Grafana to cut heap allocation by 30%.
  • Built RAG ingestion for Initializ’s AI assistants (S3, SharePoint) and an offline RAG chatbot on Ollama and Qdrant.
  • Built contract-renewal APIs in Go (Gin, MongoDB) with role-based access, plus an AI chat that answers renewal questions from live renewal, inventory and agent-activity data.

Initializ Pvt. Ltd. | Software Developer Intern

Gurugram, India | Aug 2023 – June 2024
  • Built secrets-operator, an open-source Kubernetes Operator (CRDs) for secrets; cut update latency by 67%.
  • Built the Secret Management Microservice with zero-knowledge, client-side Web Crypto encryption.

Open source

Forge | Golang, AI Agent Runtime

2026
  • Added WhatsApp support to Forge, an open-source runtime for AI agents (6,200+ lines of Go, PR #452): QR-code login, sender allow-listing, reply-loop protection and 16 end-to-end tests.

Projects

PDFFusion | Python, FastAPI, AWS S3

2025 – 2026
  • Open-source microservice that merges PDFs and images from S3 presigned URLs, validated with qpdf.

Ratn Tools Suite | TypeScript, Next.js

Aug 2025 – Present
  • JSON validation and compression tools that run fully in the browser; data never leaves the device.

Education

Dr. APJ Abdul Kalam Technical University | B.Tech in CSEKanpur, India | 2020 – 2024 | 8.0 CGPA
Maharshi Patanjali Vidya Mandir | IntermediateAllahabad, India | 2017 – 2019 | 84.02%