# Nikhil Sahu

> I build backend systems, cloud infrastructure and AI-powered products that scale, ship, and stay up. Terraform-first AWS, FastAPI/Go services, LGTM observability and production RAG are my daily tools.

Nikhil Kumar Sahu is a software engineer (backend, cloud infrastructure, distributed systems, applied AI) based in Mumbai, India, working remotely. This site is both a freelance/consulting front and an engineering portfolio. Contact: hi@nikhilsahu.dev. Book a call: https://cal.com/nikhilsahu/15min.

## Pages

- [Home / services](https://nikhilsahu.vercel.app/): what I solve — cloud & infrastructure, backend & APIs, AI/RAG/agents, observability & SRE, full-stack delivery
- [Case studies](https://nikhilsahu.vercel.app/work): detailed project write-ups with architecture and results
- [Engineering profile](https://nikhilsahu.vercel.app/portfolio): experience, projects, skills, education, certifications
- [Hiring for AI / agents](https://nikhilsahu.vercel.app/ai) · [Infra / DevOps / SRE](https://nikhilsahu.vercel.app/infra) · [Full-stack SDE](https://nikhilsahu.vercel.app/sde) · [Backend](https://nikhilsahu.vercel.app/be): role-specific summaries with FAQs
- [Blog](https://nikhilsahu.vercel.app/blog) · [Knowledge graph](https://nikhilsahu.vercel.app/graph) · [Bookshelf](https://nikhilsahu.vercel.app/bookshelf) · [Papers](https://nikhilsahu.vercel.app/papers) · [Gallery](https://nikhilsahu.vercel.app/gallery)
- [Contact](https://nikhilsahu.vercel.app/contact) · [Free architecture audit](https://nikhilsahu.vercel.app/audit) · [Newsletter](https://nikhilsahu.vercel.app/newsletter)
- Feeds: [RSS](https://nikhilsahu.vercel.app/feed.xml) · [Atom](https://nikhilsahu.vercel.app/atom.xml) · [JSON](https://nikhilsahu.vercel.app/feed.json) · Full text for LLMs: [llms-full.txt](https://nikhilsahu.vercel.app/llms-full.txt)

## Case studies

- [lgtm-sidecar: a resource-capped observability stack for small teams](https://nikhilsahu.vercel.app/work/lgtm-sidecar): ChallengeSmall teams either pay a SaaS observability bill that scales with their success, or install Grafana on the app server and watch it eat the memory the…
- [infra-ops: one Terraform + Ansible platform for dev, staging and prod](https://nikhilsahu.vercel.app/work/infra-ops): ChallengeWhen I joined, each environment had been built by hand at a different time by a different person. Staging didn't match prod, IAM users had access keys…
- [CaseCloud: multi-tenant legal SaaS with RAG over 80M+ judgments](https://nikhilsahu.vercel.app/work/casecloud): ChallengeLaw firms run on deadlines, documents and trust. CaseCloud needed three things at once:Hard isolation between firms. A missed WHERE tenant_id = ? is a…
- [Prompt-to-app BaaS: generating full-stack stores on AWS, then cutting the bill by 76%](https://nikhilsahu.vercel.app/work/prompt-to-app-baas): ChallengeThe product generates a complete e-commerce app from a prompt: REST APIs, a database schema and a frontend, deployed for the customer. The first…
- [Vedico: realtime fleet tracking for telecom field operations](https://nikhilsahu.vercel.app/work/vedico): ChallengeVedico maintains telecom infrastructure across regions ("circles"). Technicians move between towers all day; operations managers in a NOC need to know…
- [Immigration Copilot: agents that fill government forms in a real browser](https://nikhilsahu.vercel.app/work/immigration-copilot): ChallengeImmigration forms are long, stateful and unforgiving: a DS-160 session times out, pages depend on earlier answers, and applicants arrive with a folder…
- [A sub-100ms personalised surge-pricing engine for a US food marketplace](https://nikhilsahu.vercel.app/work/surge-pricing-engine): ChallengePrices were a table edited by hand. Demand spikes went unpriced, loyal customers paid the same as first-timers, and every pricing tweak was a deploy…

## Blog posts

- [Advanced RAG Architecture: Graph, Vector, Redis, and PostgreSQL](https://nikhilsahu.vercel.app/blog/advanced-rag-graph-vector-redis-postgres): Combining Graph databases, Vector databases with Reciprocal Rank Fusion (RRF), Redis caching, and PostgreSQL to build a state-of-the-art RAG system.
- [The Evolution of AI: From RAGs to Autonomous Agents](https://nikhilsahu.vercel.app/blog/ai-agents-and-rags): Exploring the transition from Retrieval-Augmented Generation (RAG) to fully autonomous AI agents capable of complex reasoning.
- [Building a Digital Garden with Next.js and Obsidian](https://nikhilsahu.vercel.app/blog/digital-garden): How this site turns an Obsidian vault into a blog, a knowledge graph and an AI knowledge base, and a tour of everything a note can contain.
- [Harnessing Agentic AI for Software Development](https://nikhilsahu.vercel.app/blog/harness-in-agentic-ai): How Agentic AI is transforming the software development lifecycle, from automated coding to complex problem solving.
- [Designing Multi-Tenant Database Architectures](https://nikhilsahu.vercel.app/blog/multi-tenant-architecture-database): A comprehensive guide to the different approaches for building multi-tenant databases in SaaS applications.
- [Infrastructure as Code: Bridging Terraform and Ansible](https://nikhilsahu.vercel.app/blog/terraform-and-ansible): A deep dive into how Terraform and Ansible complement each other in modern infrastructure provisioning and configuration management.
- [Zero long-lived cloud keys: GitHub Actions → AWS with OIDC](https://nikhilsahu.vercel.app/blog/zero-long-lived-keys-github-actions-oidc): How to delete every AWS access key from your CI and your laptop, replace them with short-lived roles, and make the blast radius of a leaked secret exactly zero.
- [Terraform drift detection that opens GitHub issues](https://nikhilsahu.vercel.app/blog/terraform-drift-detection-that-opens-issues): Drift is not a Terraform problem, it's a feedback-loop problem. Here is the nightly job that re-plans every environment and files an issue when reality disagrees with code.
- [Sub-millisecond WebSocket fan-out in Go: the Vedico realtime service](https://nikhilsahu.vercel.app/blog/go-websocket-fanout-sub-millisecond): How a single Go service pushes GPS telemetry from hundreds of field technicians to NOC dashboards in under a millisecond, with back-pressure, per-region isolation and no shared locks on the hot path.

## Bookshelf

- [Building Microservices (2nd edition)](https://nikhilsahu.vercel.app/bookshelf/building-microservices): Reading it slowly, chapter by chapter, against the two monolith splits I've already done. Mostly it confirms; occasionally it corrects.
- [Site Reliability Engineering: How Google Runs Production Systems](https://nikhilsahu.vercel.app/bookshelf/site-reliability-engineering): The chapters on SLOs, alerting and toil are the spine of how I set up on-call for small teams; the rest is Google-scale context.
- [Terraform: Up & Running (3rd edition)](https://nikhilsahu.vercel.app/bookshelf/terraform-up-and-running): The book that turned my Terraform from scripts into a platform. Most of infra-ops is this book plus scars.
- [Designing Data-Intensive Applications](https://nikhilsahu.vercel.app/bookshelf/designing-data-intensive-applications): The book I reread before every schema or messaging decision. My notes map its chapters to systems I've actually run.

## Papers

- [In Search of an Understandable Consensus Algorithm (Raft)](https://nikhilsahu.vercel.app/papers/raft-consensus): The paper that made consensus legible. I reread it whenever I need to reason about etcd behaviour during a Kubernetes control-plane incident.
- [Dynamo: Amazon's Highly Available Key-value Store](https://nikhilsahu.vercel.app/papers/dynamo): Read it to understand why DynamoDB partitions behave the way they do, and why an eventually consistent read is a feature with a price tag.
- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://nikhilsahu.vercel.app/papers/retrieval-augmented-generation): The origin of the acronym. Useful mostly as a baseline to measure how far production RAG has moved: hybrid retrieval, reranking, citations and evaluation sets.

## Experience

- Software Engineer, RaghuvarTech (Raghuvar IT Solutions LLP) (2025-11 → present): Own infrastructure, CI/CD and observability for a multi-service platform; lead architecture reviews across backend, infra and agentic-AI systems.
- Software Development Engineer, InkHub.in / BRMH.ai (2025-06 → 2025-10): Built an AI-powered backend-as-a-service that generates full-stack e-commerce apps from prompts, and re-architected its AWS footprint.
- Software Development Engineer (Remote), EatTogether & Co. (2024-11 → 2025-05): Backend engineer on a US food-tech team: real-time pricing, monolith-to-microservices refactor, cross-timezone agile delivery.

## Skills

- Backend & APIs: FastAPI, Node.js, Express.js, Go, GraphQL, REST, gRPC, WebSockets, SSE, OAuth, Celery
- Databases & Storage: PostgreSQL, PostGIS, Redis, Milvus, MongoDB, DynamoDB, MinIO, S3
- Applied AI & Search: RAG pipelines, Hybrid search (dense+sparse), RRF reranking, LangChain, Agentic AI, LLM integration (Gemini, OpenRouter, Claude), MCP, FAISS
- Containers & Orchestration: Docker, ECS, Kubernetes, Helm, ArgoCD (GitOps), Kustomize, Istio, Vault, External Secrets Operator
- Cloud (AWS) & IaC: EC2, ECS, ECR, S3, RDS, Lambda, Step Functions, SSM, IAM/OIDC, Terraform, Ansible
- CI/CD & DevOps: GitHub Actions, Environment-gated promotions, Drift detection, Automated testing, Code review practices
- Observability & Messaging: Grafana, Prometheus, Loki, Tempo, Pyroscope, OpenTelemetry (Alloy), Alertmanager, Apache Kafka
- Languages & Frontend: Python, Go, TypeScript, JavaScript, C++, SQL, Next.js, React, Tailwind CSS
- Engineering practice: System design, Microservices & distributed systems, Multi-tenant architecture, SRE, DevSecOps, RBAC, Agile/Scrum