Krishna Goswami, backend developer

Hi! I'm Krishna Goswami, a Backend Developer.

Building
scalable
systems
Backend developer specializing in building scalable and efficient systems.
PythonReactGolangDockerPostgreSQLAPIs

My Professional Journey

Building expertise through diverse experiences and challenging projects

Evolv logo

Full Stack Developer

Evolv, United Kingdom (Remote)

getevolv.ai

September 2025 - Present

  • Built the production frontend in Next.js and TypeScript with a reusable, component-driven design system.
  • Implemented GSAP and Lenis for smooth, high-performance scroll and motion interactions.
  • Own end-to-end SEO, including technical audits, structured data, and Core Web Vitals optimisation.
  • Drive GEO to improve brand visibility and citability across AI search platforms.
  • Develop internal MCP servers to expose tooling to LLM agents.
  • Maintain Playwright end-to-end test suites gating every release in CI.
TypeScriptReactNext.jsGSAPLenisSEOGEOMCPPlaywright
Summer of Bitcoin logo

Contributor

Summer of Bitcoin (LND)

April 2025 - July 2025

Implementing BIPs (Bitcoin Improvement Protocol) into Lightning Network for faster transactions (up to 5x). Developed an automated fuzzing tool for LND that uncovered 15+ backend vulnerabilities, including XSS and transaction malleability, improving test coverage by 40%.

GolangLightning Network Daemon (LND)Bitcoin CorelibFuzzergRPCXSS DetectionZero-Copy ParsingP2P Protocols

Backend Developer

Freelancer

July 2023 - Present

Developed a few production-grade backend systems using Golang, Python, and Generative Ai for overseas clients. Focused on building scalable and efficient APIs, integrating with various databases, and ensuring high availability.

API OptimizationPythonGolangGenerative AIDatabase ManagementScalability

Projects

Farmingo (NASA Space App Challenge 2024)

Farmingo (NASA Space App Challenge 2024)

  • Earth observation dashboard for farmers, built around NASA GRACE groundwater and soil moisture data to flag drought and flood risk before a season goes wrong.
  • A CNN classifier handles crop disease prediction from leaf images, while Prophet and Random Forest models forecast water demand and drive irrigation scheduling.
  • React, Tailwind, Leaflet and Chart.js on the front, with Node/Express, Django, FastAPI and Celery workers behind it, over MongoDB Atlas, Firestore and BigQuery.
GitHub
Chat App

Chat App

  • Real-time messaging over Socket.IO WebSockets, covering both direct chats and community rooms, so delivery and receipts are pushed rather than polled.
  • File and image sharing through Multer multipart uploads, with bcrypt password hashing and JWT sessions on a Node and Express backend.
  • React client using Zustand for global state, Axios for the REST calls, and Tailwind with shadcn/ui for the responsive layout.
GitHub
FactorBoard (AI solution for Whiteboards)

FactorBoard (AI solution for Whiteboards)

  • An HTML5 canvas whiteboard that reads what you draw and solves it, the way the Notes app does maths on an iPad.
  • The canvas is serialised to a base64 PNG and sent to a FastAPI service that prompts Gemini 1.5 Flash for vision inference, returning the expression in LaTeX along with the computed result and any variable bindings.
  • React and TypeScript on Vite, with Mantine, Radix primitives, Tailwind and tesseract.js for client-side OCR.
GitHub
Dynamic Multimodal RAG System

Dynamic Multimodal RAG System

  • A multimodal RAG service that ingests PDFs, CSVs and scraped web pages, chunks them, and keeps embeddings in a Chroma vector store for semantic retrieval.
  • LangChain orchestrates the OpenAI calls with per-route prompt templates, a dedicated table reasoner for structured data, and LangSmith tracing; dynamic re-indexing keeps answers current as sources change.
  • FastAPI backend with JWT auth, async MongoDB through Motor, a Dockerfile for deployment, and a plugin API so a team can drop the chat into its own product.
GitHub
ExDate

ExDate

  • A dividend calendar for retail investors. It tracks upcoming ex-dividend dates so you can see which stocks are paying out and when you need to hold by to qualify.
  • A Go scraper pulls the NSE corporate actions feed on a cron schedule, filters for equity dividend records, and writes them to Postgres through pgx.
  • A Gin REST API serves that data to a Next.js App Router frontend in TypeScript and Tailwind v4, with Vercel analytics and speed insights wired in.
Try now
SmartTree (Government of India)

SmartTree (Government of India)

  • A field cataloguing tool for a tree census: upload an XLSX or CSV inventory and get back a scannable QR label and a printable record sheet for every tree.
  • QR generation uses segno and Pillow, encoding a deep link to each tree's own PDF, and ships the whole batch as a ZIP.
  • The PDF pipeline runs on PyMuPDF, prepending a header and data table and appending a footer per record, with Devanagari font embedding for vernacular names. Streamlit handles the UI.
Try now

Tech Stack

Interface

React
Next.js
TypeScript
JavaScript
Tailwind
Redux
GSAP
Lenis

Services

Node.js
Express
FastAPI
Go
Python
Docker

Data

Postgres
MySQL
MongoDB
Redis

AI

LangChain
LangGraph
MCP

Writing

Notes on backend systems, distributed infrastructure, and things I'm learning along the way.

Read all postsShowing 2 of 2

Contact

Feel free to reach out to me via email or the contact form below. I'm always excited to connect and collaborate!

goswamikrishna549@gmail.com