Skip to content
Available for ML & full-stack roles AI/ML ENGINEER · FULL-STACK DEVELOPER

Deepak Kumar

I build AI that ships — from notebook to production, for products people use every day.

SCROLL TO EXPLORE
3+years building production systems
12shipped projects
4regions served at GraniteShares
38+automated QA scripts owned
$ python embed.py --view 3d

My work, as an embedding space.

Every project and post, embedded and mapped in 3D. Run a query to watch retrieval happen — the same idea that powers the RAG systems I build for teams.

  • Each dot is a chunk of a project or post.
  • Colours are topic clusters: LLMs & RAG, agents, full-stack and DevOps.
  • Lines link your query to its nearest neighbours.
embeddings.3dmy work as vectors · drag to rotate
The 3D view needs WebGL — try another browser.
RUN A QUERY
const developer = {
  name: "Deepak Kumar",
  role: "Full-Stack Developer · AI/ML Engineer",
  stack: ["Python", "PHP", "React/Next.js"],
  focus: ["RAG systems", "Agentic AI", "Full-stack"],
  coffee: Infinity,
  openToWork: true,
};

// ideas in, products out
export default ship(developer);
main · 0 problemsTypeScript · UTF-8 · Ln 11
$ git log --stat --since=1.year

Writes code daily. Ships weekly.

Production fixes, automation scripts and side projects, most weeks of the year. Switch the editor to attention.py and click a word.

Less More
$ cat ARCHITECTURE.md

How the systems are actually built.

The flow most of my backend work follows: Client → API Gateway → stateless Laravel/FastAPI services → distributed cache → MySQL/Postgres (primary + read replicas) → Elasticsearch/pgvector → async queue workers.

01 / 08

Caching strategy

Route-level caching with TTL, versioned cache keys, invalidation on write events, stampede protection, and cache warming for high-traffic endpoints.

Stampede-protected, TTL-versioned
02 / 08

Event-driven search

Database create/update/delete events trigger async queue jobs that update Elasticsearch, keeping it eventually consistent with MySQL — with filtering, sorting and geo-distance queries, and a DB fallback if search is down.

Eventually consistent, DB-fallback safe
03 / 08

Scalability

Stateless services that scale horizontally, MySQL read replicas for high-read workloads, and Elasticsearch index lifecycle management for large datasets.

Horizontally stateless
04 / 08

Reliability & ops

Logging, monitoring and alerting for queue lag, cache efficiency and search performance; zero-downtime deployments with rollback strategies and automated recovery.

Zero-downtime rollbacks
05 / 08 AI-NATIVE

AI in the backend

LLMs integrated as just another backend dependency — structured JSON extraction, confidence-based escalation, and real-time streaming, wired into the same queues and caches as everything else.

Confidence-gated, fully logged
06 / 08

Cost & performance

Consolidated a split deployment onto one domain, cutting hosting cost roughly 50% — along with caching, indexing and query tuning that cut response times in several optimised flows.

~50% hosting cost cut
07 / 08 SAFETY-FIRST AI

Human-in-the-loop AI

Every AI suggestion sits behind a confirmation gate, enforced in code — a teacher finalises a grade, an annotator confirms a tag. High confidence never means automatic.

0 auto-approved grades
08 / 08

Local-first when it matters

Not everything needs the cloud — Ermis V2 runs entirely on-device (local Whisper, SQLite, zero telemetry) for work where the data can never leave the machine.

0% data leaves the device
$ ./services --list

What I can build for you.

Whether you’re hiring for your team or need a product shipped, these are the problems I solve end to end — from the first prototype to monitoring in production.

Discuss a project
01 / 06 AI-NATIVE

LLM apps & RAG

Assistants that answer from your own documents, agents with guardrails, and search that understands meaning.

  • Retrieval pipeline with citations
  • Evaluation suite & quality dashboard
  • Production API with cost controls
02 / 06

Full-stack platforms

CRMs, booking platforms and fleet/ops tools — built end to end on Laravel, React or Vue, with real cross-platform (Android/Electron) support.

  • Schema design & REST APIs
  • React.js / Vue.js frontends
  • Payments, auth & reporting
03 / 06

DevOps & CI/CD automation

The plumbing that keeps a team's pipelines honest: automated QA suites, serverless alerting and schedulers that replace fragile crontabs.

  • Azure DevOps / GitHub Actions pipelines
  • Automated QA & data-quality checks
  • Monitoring, alerts & secret hygiene
04 / 06

Full-stack AI products

The product around the model: fast interfaces, solid APIs, auth and analytics — so users actually adopt it.

  • Next.js frontend & design system
  • FastAPI / Node backend
  • Auth, analytics & feedback loops
Proof: MailAI →
05 / 06 LOCAL AI

Desktop & offline AI tools

Local-first apps that use AI without sending data anywhere — on-device speech, vision or NLP, for work that can't touch the cloud.

  • PySide6 / desktop UI
  • On-device ASR & rule engines
  • SQLite audit trails, zero telemetry
06 / 06

Data pipelines & automation

ETL pipelines that turn messy, unstructured sources into validated, structured data at scale, with checkpoint recovery built in.

  • Checkpointed ETL & batch processing
  • Schema validation & data QA
  • Scheduled, zero-downtime deploys
$ ls ~/projects --featured

Built, shipped, measured.

All projects
~/projects/fleet-tracking

Driver & Fleet Tracking App

Real-time GPS tracking over WebSocket with OpenStreetMap routing and a dispatcher desktop companion.

Vue.jsWebSocket · OpenStreetMapCase study
~/projects/gscheck-financial-qa

GScheck — Financial Data QA

38+ script automated QA suite across 4 regions, with an 8-trigger Azure Pipelines scheduler and serverless alerting.

PythonAzure DevOpsCase study
~/projects/ai-data-extraction-seo

AI Data Extraction & SEO Platform

Distributed ETL pipeline turning unstructured HTML into a validated 25+ field JSON schema, 100+ records/day.

PythonNVIDIA NIM · PostgreSQLCase study
~/projects/ermis-v2-speech-assistant

Ermis V2 — Speech Annotation Assistant

Local desktop app: on-device Whisper drafting, a 26-rule compliance validator, and an append-only audit log. 100% local, no cloud.

PythonPySide6 · SQLiteCase study
~/projects/mailai

MailAI — AI-Powered Email Client

Gemini AI drives the whole UI by natural language — compose, send, navigate, filter. Built in a 5-day hiring assessment.

React 18Gmail API · OAuth 2.0Case study
~/projects/eventflub

EventFlub.com — Event Booking Platform

Built from scratch, solo: real-time availability, payment-gated booking confirmation, full Laravel backend.

LaravelMySQL · PaymentsCase study
~/projects/ai-web-search-engine

AI-Powered Web Search Engine

Multi-engine aggregator with Llama 3.3 content extraction, dual-layer caching, and real-time SSE streaming.

FlaskLlama 3.3 · SSECase study
$ cat workflow.yaml

How I work.

A simple loop that keeps ML projects honest: agree on what “good” means, get something working fast, measure it, then ship and keep watching.

  1. 01

    Understand

    Talk to users, map the data, and pick the one metric that proves the project worked.

    output: success metric + plan
  2. 02

    Prototype

    A working baseline in days, not months — real data, simple model, something people can click.

    output: clickable demo
  3. 03

    Evaluate

    Test sets, not vibes. Every change is scored for quality, latency and cost before it moves forward.

    output: eval report
  4. 04

    Ship & monitor

    Deploy behind a flag, watch real usage, and keep improving with feedback from production.

    output: live product + dashboard
$ jupyter lab

Where ideas get tested first.

Fine-tunes, notebooks, live demos and paper breakdowns — small experiments, kept in the open.

Open the full lab log
package.json — what I work with
"dependencies": {
  "python": "daily",
  "php": "daily",
  "langchain": "production",
  "hugging-face": "production",
  "fastapi": "production",
  "laravel": "production",
  "pgvector": "production",
  "next.js": "production",
  "azure-devops": "production",
  "go": "production",
  "kubernetes": "learning"
}
lab.ipynb
In [1]:
FINE-TUNEHugging Face LLM fine-tuningFine-tuned a model to reduce hallucinations in the AI Teacher Assistant's generated answers.
In [2]:
VISIONPedestrian Detection & Gender ClassificationReal-time detection and classification on video frames with YOLOv8 and OpenCV.
In [3]:
ASSISTANTVirtual Personal AssistantVoice recognition, text-to-speech and NLP-driven task automation with the OpenAI API.
In [4]:
WRITE-UPInside the AI Teacher AssistantQuestion papers, answer evaluation and why the AI never finalises a grade.
$ ls ~/stack --format=globe

A stack built for shipping production AI.

The 37 tools I reach for across backend, AI/LLM, frontend and data — from a Laravel API to a RAG pipeline to the Postgres behind it. Pick a layer to light it up on the globe — or drag it to spin.

stack.globedrag to spin
The 3D view needs WebGL — the full list is on the left.
$ cat now.md

What I’m up to now.

BUILDING
Just shipped Ermis V2 (1,552 tests passing) — now back to production RAG and agentic AI systems
LEARNING
Financial-market systems from GraniteShares' ETF/Bloomberg work, plus Kubernetes and GCP
OPEN TO
Full-stack and AI/ML engineering roles, remote-friendly
WRITING
Next post: migrating pipeline alerting off SMTP to serverless
BASED IN
New Delhi, India
Updated Oct 2026
$ cat about.md

Curious engineer. Practical builder.

DK

I enjoy the whole path from a notebook experiment to a reliable product — owning requirements, design, build, testing and deployment myself on most of the roles I've worked.

Right now I'm focused on production RAG systems, agentic workflows and full-stack delivery. Open to full-stack and AI/ML engineering roles, ideally on product-focused teams.

Download résumé
git log --graph --oneline career
a3f9c21HEAD → mainAug 2026 — now

feat: Software Engineer at GraniteShares

Own the public website (ASP.NET/C#/SQL Server/Umbraco) and built GScheck — a 38+ script automated QA suite across 4 regions, plus the Azure CI/CD behind it.

7b2e0d4Feb 2025 — Apr 2026

feat: Sr. Software Developer at Bharat Software Solutions

Sole engineer on a multi-tenant Laravel ERP + FastAPI AI service: RAG-grounded lesson planning, AI question-paper generation with a leak-protected per-school bank, and a rule-engine + LLM hybrid for answer evaluation. Also built the fleet/GPS tracking platform and a custom CRM.

19c4e8aJun 2024 — Feb 2025

feat: Sr. Software Developer at Comtech Training & Technologies

Migrated search to Elasticsearch across two production sites (custom analyzers, synonyms, field boosts), with blue/green index rollover for zero downtime.

5a1c8f0Dec 2023 — Jun 2024

feat: Software Developer at Picera Private Limited

CQRS-style CRM: separated command/query paths, precomputed dashboard aggregates to a Redis-like cache, Elasticsearch fuzzy search on leads, and async CSV export via pre-signed S3 URLs.

c2d7e91Jun 2023 — Nov 2023

feat: PHP Web Developer at SS Agriculture Innovation (remote)

E-commerce flows with cached catalog read models and a strictly ACID checkout path using the outbox pattern for notifications and invoice PDFs; seek-based pagination for large lists.

0d1f2b32023 — 2027

init: B.Tech Computer Science at AKTU Lucknow

$ grep -r "worked with" ./reviews

Kind words.

Testimonials coming soon.
$ ls -t ~/blog | head -4

Fresh from the blog.

All posts
$ ./contact.sh

Let’s build something worth shipping.

Hiring for an ML or full-stack role, exploring a collaboration, or just want to talk models? My inbox is open.

EMAILdk78834@gmail.com
contact.sh — zsh80×24
Ask my AI twindemo · answers from this site