Prashant Kumar

Prashant Kumar

+•• ••••• •••••••••••@•••••.com

6.6 yrs

As an independent full stack developer, I am driven by a deep desire for autonomy and mastery, meticulously crafting solutions with a logical and pragmatic mindset. I thrive on the creative challenge of architecting robust systems, always seeking to innovate and achieve excellence in every project I undertake.

  • Delhi
  • Remote

Ask about Prashant

Skills

Claimed

Tool DevelopmentAccount ManagementTime ManagementDue DiligenceRequirements AnalysisWork Planning and PrioritizationStructured Work ApproachProblem SolvingMentoring and Peer CoachingUser FeedbackProduct DemonstrationsMeeting Scheduling

Proof of Experience

Independent Full Stack Developer

US Client · Jun 2026 – Jul 2026

  • Delivered an end-to-end SaaS product on Next.js, Supabase with Drizzle ORM and Stripe - owning requirements breakdown, technical documentation, implementation and deployment.
  • Built an AI orchestration layer generating individual and compatibility reports from user inputs, plus a Content Library to manage ingested questions and the prompts behind each report.
  • Enforced multi-tenant data security with PostgreSQL Row Level Security and Supabase Auth.
Edvanta Technologies

Full Stack Developer

Edvanta Technologies · Dec 2020 – Apr 2026

  • Core engineer on ThoughtJumper, a knowledge-graph learning platform - built the main Vue.js workspace around an interactive AntV G6 canvas, with a type-driven widget system rendering text, audio and video nodes through separate components, backed by Node.js/Express REST endpoints and MongoDB schemas.
  • Redesigned the graph data model - split one monolithic MongoDB document into separate Node and Edge collections - so a single node edit stopped rewriting the entire graph, removing write amplification as graphs grew.
  • Cut API response times by 90% (250ms 20ms) and Redis cache footprint by 98% (2.5MB 50KB) by gzip-compressing graph payloads before caching.
  • Evolved the knowledge source layer across three generations - Wikipedia scraping, then structured entity extraction from Wikidata/DBpedia, then LLM-generated nodes and edges - mapping entities into the graph schema at each stage, with a batch enrichment path that populates blank nodes in groups instead of per-node API calls.
  • Implemented a Graph-RAG pipeline (LangChain + Pinecone) that ingests each user's graph into its own vector index, referenced from graph metadata, so every user queries only their own knowledge base.
  • Built a companion Chrome extension (React + Webpack) that captures page content via context menu into a dedicated Node.js event service, which persists the node and pushes a live, user-scoped update to that person's open graph.
  • Developed a real-time collaborative graph canvas with Socket.IO, letting multiple concurrent users edit and visualize a shared graph.
  • Built a career-profiling product (React/Node.js) that parses uploaded resumes, matches them against aggregated job listings and surfaces per-role skill gaps - linking each missing skill back to the knowledge-graph platform to learn it.
  • Also built its market-insights dashboard, ingesting US Bureau of Labor Statistics data into MongoDB to compare employment trends across states.

Technical Blog

LogChit · Oct 2019 – Apr 2023

  • Experiment and Knowledge Sharing

Proof-of-Work

No completed sessions yet. Anything Prashant finishes shows up here with the evidence behind it.

Proof of Education

Bachelor, Computer Applications

Maharshi Dayanand University, Rohtak · 2017 — 2020

A Proof CV — one profile, kept current, with the proof attached. Back to the portal