DK / v4.26
BLR--:--:--IST

Issue N°04 — Portfolio & Field Notes / 2026

Dheerendra
Kumar.

I build the frontend layer of enterprise software — platforms, design systems, and AI-native workflows that ship to real customers and move real revenue.

→ 4+ yrs shipping product
→ React · TS · Node · Python
→ LLM / RAG / Vector DBs
→ Currently: Bengaluru, IN
Frontend Architecture · Platform Engineering · Design Systems · AI Systems · LLMs · RAG · Storybook · Vite · React · Next.js · TypeScript · Frontend Architecture · Platform Engineering · Design Systems · AI Systems · LLMs · RAG · Storybook · Vite · React · Next.js · TypeScript · Frontend Architecture · Platform Engineering · Design Systems · AI Systems · LLMs · RAG · Storybook · Vite · React · Next.js · TypeScript · Frontend Architecture · Platform Engineering · Design Systems · AI Systems · LLMs · RAG · Storybook · Vite · React · Next.js · TypeScript · Frontend Architecture · Platform Engineering · Design Systems · AI Systems · LLMs · RAG · Storybook · Vite · React · Next.js · TypeScript · Frontend Architecture · Platform Engineering · Design Systems · AI Systems · LLMs · RAG · Storybook · Vite · React · Next.js · TypeScript · Frontend Architecture · Platform Engineering · Design Systems · AI Systems · LLMs · RAG · Storybook · Vite · React · Next.js · TypeScript · Frontend Architecture · Platform Engineering · Design Systems · AI Systems · LLMs · RAG · Storybook · Vite · React · Next.js · TypeScript ·
$500K → $2M+
ARR growth during my tenure at Peoplebox
34 → 6 min
Deployment time after CRA → Vite migration
~20%
Product revenue attributed to owned modules
4+ yrs
Shipping enterprise frontend at scale
§ 01 — Field Work

Where I've built things

— scroll ↓
#01
May 2025 — Present
Bengaluru

Peoplebox

YC W22

Senior Software Engineer

  • ▍Owned end-to-end development of Goals, Team Management, Continuous Feedback & Reporting modules — from discovery through production.
  • ▍Led frontend architecture for the Platform team; shipped Storybook as the shared design system and set engineering standards.
  • ▍Drove the CRA → Vite migration. Deployment time fell from 34 min to 5–6 min.
  • ▍Built Microsoft Teams integration, AI-powered recruitment workflows & customizable reporting used by large orgs.
  • ▍During tenure, company scaled ARR from ~$500K to $2M+ — owned modules accounted for ~20% of product revenue.
#02
Sep 2024 — Apr 2025
Bengaluru

Peoplebox

YC W22

Software Engineer

  • ▍Delivered customer-facing features across the full development lifecycle with cross-functional teams.
  • ▍Built reusable components, reporting tools, dashboards & internal platform features that improved scalability.
#03
Feb 2024 — Jun 2024
Remote

Onito Technologies

Startup

Software Engineer

  • ▍Shipped a pharmacy feature improving accessibility for healthcare professionals.
  • ▍Migrated codebase from React to Next.js — measurable improvements in load times & scalability.
  • ▍Integrated Google BigQuery + Python backend to store, retrieve & analyze large datasets.
#04
Aug 2021 — Aug 2023
Remote

Spacephilic

Frontend Engineer

  • ▍Implemented real-time data transmission & calculations for accurate live information.
  • ▍Used Firebase to sync user data seamlessly across web and mobile.
§ 02 — Toolbelt

The stack

— scroll ↓
Languages
  • TypeScript
  • JavaScript
  • Python
  • HTML
  • CSS
Frameworks
  • React
  • Next.js
  • Node.js
  • Express
  • Tailwind
  • MUI
  • Bootstrap
Data
  • MongoDB
  • BigQuery
  • Firebase
  • Vector DBs
AI / ML
  • LLM Integration
  • RAG
  • Fine-tuning
  • Embeddings
Tooling
  • Vite
  • Docker
  • AWS
  • Vercel
  • Capacitor
  • Android Studio
  • GitHub
§ 03 — From the lab

Side experiments

— scroll ↓
exp/001 · Aug 2024
Python · FFmpeg · Embeddings · LLM

AI Image & Video
classification system.

An AI system that classifies short-form video and images as cringe, engaging, or high-quality. FFmpeg pulls frames, vector embeddings drive a cosine-similarity pre-filter, and only the ambiguous cases hit the LLM — cheap, fast, weirdly accurate.

Frame extractionVector embeddingsCosine pre-filterLLM fallbackPipeline
Currently exploring
Agent frameworks, evals & tight feedback loops for LLM products.
Belief
"The frontend is where product actually happens."
§ 04 — End of transmission

Let's build something.

— scroll ↓
developer.
dheerendra@gmail.com_
GitHub↗LinkedIn↗LeetCode↗Phone↗